THE IMPACT OF DEMOGRAPHIC CUES IN VIDEO-BASED SITUATIONAL JUDGMENT
ITEMS
By
Juliya Golubovich

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS
Psychology
2011

ABSTRACT
THE IMPACT OF DEMOGRAPHIC CUES IN VIDEO-BASED SITUATIONAL JUDGMENT
ITEMS
By
Juliya Golubovich
Situational judgment tests (SJTs) present test takers with hypothetical work-related
scenarios and offer options of possible responses. Such tests are often video-based; test takers
watch videos in which actors act out each scenario. This study examined the implications of
actor casting choices. A relational demography perspective, which considers how demographic
similarity between individuals influences the quality of their interactions with one another and
their individual outcomes, was applied to examine the effects of actor demographics in a videobased SJT on test takers’ responses and reactions. The race and gender of the actor participants
were asked to imagine interacting with were varied to film six different versions of each of four
SJT scenarios. Participants viewed one of the six versions of the test and indicated their
likelihood of responding to each scenario in a number of different ways. Participants also
answered questions about their reactions to the test. Findings showed that test takers did seem to
attend to the race and gender of those they ―interacted with‖ in SJT scenarios and this influenced
their responses to and actual performance on the test. Actor demographics had little influence on
test takers’ reactions to the test, however. Implications of the findings and limitations are
discussed.

ACKNOWLEDGMENTS

There is a whole army of wonderful people who helped to make this project possible and
it gives me great pleasure to have this opportunity to thank them. First of all, I would like to
thank my advisor, Dr. Ann Marie Ryan, for her steadfast guidance, timely feedback, and
encouragement throughout this process. I would also like to thank my other committee members,
Dr. Neal Schmitt and Dr. John Schaubroeck for their critical feedback and helpful insights. I
would like to express my gratitude to Proctor & Gamble, and particularly Dr. Robert E. Gibby,
for offering the financial support that made this study possible.
I am also extremely grateful to the 28 individuals who acted in my videos for responding
to my call for help and doing an excellent job, and to the faculty members, students, and friends
who helped me seek them out. I also want to thank the forty individuals who served as judges of
my situational judgment items and actively helped me recruit additional people for the task.
I would like to acknowledge the team at WKAR, Michigan State University’s on-campus
TV station, for their professional directing, filming, and production work and Gary Schrock for
his technical assistance in getting all the videos reformatted and set up for the experiment. A
special thanks to my exceptional team of research assistants, Jessie Olien, Abraham Kim, and
Julie Klopp, who I could always depend on two hundred percent during the data collection
process, and to Leslie Baldwin for her speedy and always friendly assistance with all issues
related to the participant pool.
I want to offer heartfelt thanks to my wonderful mom and older brother for their
continuous moral support and earnest attempts to understand what I was trying to do. Thank you,
mom, for the 24 pound care packages which had the UPS guy huffing and puffing on the way to
my door.
iii

Finally, thank you Alex, for listening patiently to my kvetching, ordering me takeout
when my food supplies dwindled, and for being there for me always.

iv

TABLE OF CONTENTS

LIST OF TABLES ...................................................................................................................... vii
LIST OF FIGURES ..................................................................................................................... ix
INTRODUCTION ........................................................................................................................ 1
Situational Judgment Tests ......................................................................................................... 3
Relational Demography ............................................................................................................ 10
Helping Situations ..................................................................................................................... 16
Conflict Situations .................................................................................................................... 27
Applicant Reactions .................................................................................................................. 48
METHOD .................................................................................................................................... 51
Design ....................................................................................................................................... 51
Sample....................................................................................................................................... 52
Procedure .................................................................................................................................. 52
Measures ................................................................................................................................... 53
RESULTS .................................................................................................................................... 60
Manipulation Checks ................................................................................................................ 60
Familiarity with Actors ............................................................................................................. 63
Item Weights ............................................................................................................................. 63
Reliability of Measurement of Hypothesized SJT Item Dimensions........................................ 73
Correlations ............................................................................................................................... 77
Tests of Hypotheses Pertaining to Helping ............................................................................... 88
Tests of Hypotheses Pertaining to Conflict Situations ........................................................... 108
Correlations with Applicant Reactions ................................................................................... 124
Tests of Hypotheses Pertaining to Applicant Reactions ......................................................... 125
Exploratory Analyses Using Overall Test Score .................................................................... 131
DISCUSSION ............................................................................................................................ 138
Study Findings ........................................................................................................................ 138
Practical Implications.............................................................................................................. 142
v

Limitations and Future Directions .......................................................................................... 143
Conclusion .............................................................................................................................. 150
APPENDICES ........................................................................................................................... 152
APPENDIX A ......................................................................................................................... 153
APPENDIX B ......................................................................................................................... 154
APPENDIX C ......................................................................................................................... 155
APPENDIX D ......................................................................................................................... 159
APPENDIX E ......................................................................................................................... 160
APPENDIX F.......................................................................................................................... 161
APPENDIX G ......................................................................................................................... 162
APPENDIX H ......................................................................................................................... 163
APPENDIX I .......................................................................................................................... 165
REFERENCES.......................................................................................................................... 167

vi

LIST OF TABLES

Table 1. Distribution of incorrect responses across manipulation checks .................................... 62
Table 2. Means, standard deviations, and rWG values for judges' ratings of conflict response
options ........................................................................................................................................... 65
Table 3. Means, standard deviations, and rWG values for judges' ratings of helping response
options ........................................................................................................................................... 66
Table 4. Means, standard deviations, and rWG values for judges' standardized ratings of conflict
response options ............................................................................................................................ 68
Table 5. Means, standard deviations, and rWG values for judges' standardized ratings of helping
response options ............................................................................................................................ 69
Table 6. Unit weights associated with judges' standardized ratings of conflict response options 71
Table 7. Unit weights associated with judges' standardized ratings of helping response options 72
Table 8. Reliabilities of SJT item dimensions .............................................................................. 74
Table 9. Loadings for likelihood ratings respondents gave to SJT response options ................... 75
Table 10. Intercorrelations, means, standard deviations, and reliabilities .................................... 79
Table 11. Likelihood ratings for helping response options as a function of respondent's and nonfocal actor's sex ............................................................................................................................. 89
Table 12. Likelihood ratings for helping response options as a function of respondent's and nonfocal actor's race............................................................................................................................ 90
Table 13. Likelihood ratings for three response options as a function of non-focal actor sex ..... 96
Table 14. Groups compared in testing Hypothesis 4 .................................................................... 98
Table 15. Observed frequencies of Black respondents' least likely responses on helping situation
2................................................................................................................................................... 104
Table 16. Summary of findings for helping situations ............................................................... 106
Table 17. Effects of demographic cues on performance on helping items ................................. 107

vii

Table 18. Likelihood ratings for conflict response options as a function of respondent's and nonfocal actor's sex ........................................................................................................................... 109
Table 19. Likelihood ratings for conflict response options as a function of respondent's and nonfocal actor's race.......................................................................................................................... 110
Table 20. Summary of findings for conflict situations ............................................................... 121
Table 21. Effects of demographic cues on performance on conflict items ................................. 123
Table 22. Reactions as a function of respondent's and non-focal actor's sex ............................. 126
Table 23. Reactions as a function of respondent's and non-focal actor's race ............................ 127
Table 24. Summary of findings for applicant reactions .............................................................. 130
Table 25. Results of ANOVA for effects of race, gender, race match, gender match, and their
interactions on test scores ........................................................................................................... 132
Table 26. Helping item 2 score differences between Black and White test takers ..................... 135
Table 27. Additional analyses of effects of demographic cues on performance ........................ 137
Table 28. Distribution of responses for those who failed the manipulation check for conflict
situation 1 .................................................................................................................................... 165
Table 29. Distribution of responses for those who failed the manipulation check for conflict
situation 2 .................................................................................................................................... 165
Table 30. Distribution of responses for those who failed the manipulation check for helping
situation 1 .................................................................................................................................... 166
Table 31. Distribution of responses for those who failed the manipulation check for helping
situation 2 .................................................................................................................................... 166

viii

LIST OF FIGURES

Figure 1. Illustration of Hypothesis 1 ........................................................................................... 20
Figure 2. Illustration of Hypothesis 3 ........................................................................................... 22
Figure 3. Illustration of Hypothesis 4 ........................................................................................... 25
Figure 4. Illustration of Hypothesis 5a ......................................................................................... 26
Figure 5. Illustration of Hypothesis 5b ......................................................................................... 27
Figure 6. Illustration of Hypothesis 6a ......................................................................................... 30
Figure 7. Illustration of Hypothesis 6b ......................................................................................... 31
Figure 8. Illustration of Hypothesis 7 ........................................................................................... 33
Figure 9. Illustration of Hypothesis 8 ........................................................................................... 34
Figure 10. Illustration of Hypothesis 10a ..................................................................................... 37
Figure 11. Illustration of Hypothesis 10b ..................................................................................... 38
Figure 12. Illustration of Hypothesis 11 ....................................................................................... 40
Figure 13. Illustration of Hypothesis 12 ....................................................................................... 41
Figure 14. Illustration of Hypothesis 13a ..................................................................................... 43
Figure 15. Illustration of Hypothesis 13b ..................................................................................... 44
Figure 16. Illustration of Hypothesis 14a ..................................................................................... 45
Figure 17. Illustration of Hypothesis 14b ..................................................................................... 46
Figure 18. Illustration of Hypothesis 15 ....................................................................................... 47
Figure 19. Illustration of Hypothesis 16 ....................................................................................... 48
Figure 20. Effects of non-focal actor race and gender on helping situation 1 score for male and
female respondents........................................................................................................................ 99
Figure 21. Effects of non-focal actor race on helping situation 2 score for different groups of
respondents ................................................................................................................................. 102
ix

Figure 22. Effects of race match with non-focal actors on test scores for different groups of
respondents ................................................................................................................................. 133
Figure 23. Effects of race match with the non-focal actor on helping situation 2 scores for
different groups of respondents .................................................................................................. 134

x

INTRODUCTION
Situational judgment tests (SJTs), which present individuals with hypothetical workrelated situations, are a popular tool in the realm of personnel selection and have a fairly long
th

history of use going back to the 19 century (Whetzel & McDaniel, 2009). As such, many issues
pertaining to their use (e.g., validity, response format, methods of scoring, subgroup differences)
have been subjected to considerable research attention. With the advent of video-based versions
(as compared to the typical paper-and-pencil format) of such instruments, however, new
questions arise and have yet to be explored. One such question, and one which to my knowledge
has not heretofore been examined, pertains to the impact that demographic cues in video-based
SJTs might have on how test takers construe the situations presented.
Social categorization is known to influence how individuals perceive others and act in
given situations (e.g., Tajfel, Billig, Bundy, & Flament, 1971; Carli, 1989). When little
information about others is available, individuals use social category information to form
impressions about these target individuals (Kunda & Thagard, 1996). Subsequent behavior
toward these targets (and more generally) is influenced by initial impressions. It is therefore
important to consider the influence of social categorization in selection settings. That is, how
does social categorization influence applicants’ perceptions and behaviors? Social categorization
may come into play when applicants view a recruitment video featuring ―employees‖ of
particular demographic backgrounds, perceive the demographics of an interviewer, take note of
the demographic characteristics of fellow applicants going through an assessment center, and
read test items with male or female names. These are just several examples of recruiting or
testing situations in which social categorization has the potential to influence applicant
perceptions and behaviors.
1

Video-based situational judgment items provide yet another context that is susceptible to
influences of social categorization. Test takers see the demographics (race, gender, age) of the
actors in the video and use this social category information to form impressions about them.
Given that responses to SJT items depend on how the items are interpreted (Brooks &
Highhouse, 2006), demographic cues of stimulus actors may affect test takers’ performance. To
the extent that respondents’ own race and gender help to determine the influence of actors’
demographics, as relational demography research would suggest they would, subgroup
differences in item responses may result. The aim of the study is to add to the research on the
influence of social categorization in selection settings by exploring the question of how the race
and gender of the actors in video-based SJT scenarios relative to the race and gender of the test
takers influence responses to these scenarios. Aside from performance, applicant reactions are
explored as well. These issues are investigated using two common types of SJT scenarios.
The study complements and informs existing research pertaining to the SJT and test bias,
and potentially offers practitioners some much-needed guidance in choosing actors for their
video-based tests. As it stands, test developers are guided in their casting choices by the desire to
project a positive (i.e., diverse) image of the hiring organization but have little to no research
data on what implications these choices have for the performance or reactions of test takers from
different groups.
I begin the paper with a brief review of existing research on paper-and-pencil based SJTs
and their more recent variant, the video-based SJT. I then introduce the relational demography
perspective, which was used as the springboard for making predictions about the effects of
actors’ demographic cues on responses of test takers from different groups, and go on to briefly
review the research on race and sex effects in the types of situations of interest in this study.

2

Findings from this situation-specific research will in some cases depart from the relational
demography perspective that I will be using as the point of departure. I will point this out where
applicable. I wrap up the theory-based portion of the paper with a brief review of the relevant
research on applicant reactions and thereafter delve into describing my method of carrying out
the study. The results of the study are then presented and the implications and limitations of the
findings are discussed. Future directions for this line of research are suggested.
Situational Judgment Tests
Situational judgment tests (SJTs) are ―measurement methods that may be used to assess a
variety of constructs‖ (McDaniel, Morgeson, Finnegan, Campion, & Braverman, 2001). SJT
items proffer hypothetical scenarios of problems that may arise in a work setting and options of
possible responses to these situations. The situations presented may reflect actual situations that
could occur on the job that the test was designed to reflect, or situations that are designed to be
psychologically analogous to actual job situations (Chan & Schmitt, 1997). The premise
underlying asking individuals to respond to hypothetical situations is that intentions predict
future behavior (Latham, Saari, Pursell, & Campion, 1980).
SJT response instructions. After each situation is presented, the examinee is asked to
indicate how he or she would respond if confronted with the given situation. Response
instructions to SJTs can have either a knowledge or a behavioral tendency format (Lievens,
Peeters, & Schollaert, 2008). With a knowledge format, examinees are asked to provide the best
answer; with a behavioral tendency format, they are asked to choose their own most likely
response. SJTs with a knowledge format are akin to cognitive ability or job knowledge tests in
that examinees are asked to show that they know what the most effective response to a particular
situation would be (Lievens et al., 2008). On the other hand, SJTs with a behavioral tendency

3

format are similar to personality tests in that examinees are instructed to demonstrate what their
own behavior would be in a given situation (Lievens et al., 2008).
SJT development process. SJT development usually involves three stages. Lievens,
Peeters, and Schollaert (2008) detail the process, drawing partly on the steps described by
Motowidlo, Dunnette, and Carter (1990) in their work. In the first step, a job analysis is
conducted, during which subject matter experts (e.g., supervisors) provide the SJT developers
with instances of critical incidents (i.e., instances of behavior that reflect effective and ineffective
performance) at work. Then, test developers proceed to group these critical incidents into
particular elements of performance, and to write descriptions of task situations based on the
incidents. In writing these situations, developers try to keep to a certain length and format. In the
second step of SJT development, a new group of subject matter experts or novice employees are
instructed to come up with possible responses for each job scenario written by the developers.
While subject matter experts are able to generate the optimal responses for the various scenarios
as well as some inferior ones, inexperienced employees can contribute alternatives along a
relatively wider continuum of effectiveness. Armed with a range of response alternatives for the
job situations, SJT developers can select for inclusion on the test response alternatives that fall
along a continuum of effectiveness. The third and final step of SJT development is the creation
of a scoring key.
SJT scoring. Scoring keys for SJTs are typically developed via one of the following
three methods: empirical, theoretical, or expert-based (Bergman et al., 2006). When SJT items
are empirically scored, a criterion measure (e.g., job performance) is chosen and items or
response options are scored according to how well these items or response options are able to
distinguish high from low scorers on the criterion measure (Lievens et al., 2008). This requires

4

giving SJT items to a sample of individuals to complete. Response options that are endorsed
most frequently by individuals exhibiting a high score on the criterion measure are scored as
correct, while options endorsed most frequently by low scorers on the criterion measure are
scored as incorrect (Lievens et al., 2008). One of the obvious challenges inherent to this method
of key development is that empirical keys are dependent upon the quality of the selected criterion
(Campbell, 1990a, b as cited by Bergman et al., 2006).
In the case of theoretical scoring of SJTs, theory serves as the basis for the writing of
items and options, or for the scoring of chosen options on a test (Bergman et al., 2006). Options
that are consistent with theory are considered correct; options that are inconsistent with theory
are considered incorrect. Mumford and colleagues (2008) used this method to create their Team
Role Test. The test was designed to assess individuals’ knowledge of team roles and of the
situations when these roles are appropriate to use. The researchers wrote scenarios relevant for
each team role (e.g., contributor, critic) defined by Mumford and colleagues (2006). Each
scenario was designed to be most amenable to assuming one role over the others. For a given
scenario, behaviors consistent with the role considered by the researchers to be most appropriate
were considered correct answers, and behaviors inconsistent with the targeted role were
considered incorrect answers. Olson-Buchanan and her colleagues (1998) created a key to score
their conflict resolution SJT using Keenan and Olson’s (1991, as cited in Olson-Buchanan et al.,
1998) model prescribing what problem-resolving actions managers should take based on three
dimensions of the conflict situation (i.e., short-term, long-term, interpersonal). Analogous to the
empirical scoring method’s reliance on criterion quality, the theoretical scoring approach must
rely on the quality of the theory used.

5

Expert-based scoring involves asking either subject matter experts to select the best and
worst response option for each item (e.g., Banki & Latham, 2010), or asking novices and experts
to select just the optimal response for each item. In the latter case, options that experts select
most often are considered correct, irrespective of how often novices choose these same options.
In the meantime, options that novices, but not experts, choose most often are considered
incorrect (Bergman et al., 2006).
An alternative to using just one of the three methods of developing scoring keys reviewed
above is to use a hybrid of two keys that have been developed separately (see Olson-Buchanan et
al., 1998 for an application of a hybrid method). This method is discussed by Bergman and her
colleagues (2006). In one of a number of possible instantiations of the hybrid approach, one key
serves as the primary key while the second key is used to remove zeros (assigned to responses
that are neither correct nor incorrect for a given item) that result from the application of the
primary key. Combinations of empirical keys with theoretical keys have the potential to address
concerns about the individual methods (i.e., insufficient or excessive reliance on either theory or
empiricism) but flaws that might be present in the separate keys are unfortunately not expunged
when keys are combined.
Mode of SJT administration. An alternative to paper and pencil SJTs are SJTs that are
video-based. In the latter case, hypothetical scenarios are presented to individuals visually, via
video. Actors act out the situations in the actual setting of the job (e.g., Weekley & Jones, 1997)
or a setting that resembles the actual setting (e.g., Dalessio, 1994). The video can be viewed by a
group of test takers simultaneously or individually on computer screens. Test takers choose a
response to each situation from a set of response options during the pause that is built into the
video following each vignette. Video-based SJTs typically incorporate into the video a narrator

6

who may provide background information, instruct test takers on how to respond to the items,
and perhaps even introduce each scenario (e.g., Dalessio, 1994; Jones & DeCotiis, 1986;
Weekley & Jones, 1997).
Video-based SJTs have a number of advantages over their written counterparts. Because
this format allows for the presentation of both verbal and visual cues (Jones & DeCotiis, 1986),
video-presented scenarios appear more detailed and realistic (Weekley & Jones, 1997). Several
benefits of video-based SJTs stem from this. First, the detail and realism of video-based SJTs
may increase the fidelity of the situational judgment test and result in higher predictive validity
of the instrument than a paper-and-pencil based equivalent might demonstrate (Weekley &
Jones, 1997). Lievens and Sackett (2006) found some evidence of this in a field study. A videobased SJT designed to assess interpersonal and communication skills demonstrated significantly
higher validity for predicting medical students’ performance in interpersonally-oriented courses
than did a written SJT.
A second advantage of video-based SJTs over their written counterparts is their higher
realism. The realism video-based SJTs offer allows job applicants to get a better sense of the
demands of the job for which they are applying. Applicants may choose to withdraw from the
application process if they discover that the job may not be a good match (Jones & DeCotiis,
1986). Realism leads to the third advantage of video-based SJTs: realism is likely to be
associated with a high level of face validity, leading applicants to react more favorably to this
medium than they might to a written version of the test (Weekley & Jones, 1997). Several studies
have found evidence of enhanced favorability in reactions to video-based SJTs. Chan and
Schmitt (1997) administered a situational judgment test of work habits and interpersonal skills to
undergraduate students in a lab setting. Students took either a paper-and-pencil or video-based

7

format of the test. The researchers found that students perceived the video-based format to have
higher face validity with respect to a specific job they were asked to consider. In a later study,
Richman-Hirsch, Olson-Buchanan, and Drasgow (2000) administered one of three different
formats of a situational judgment test of conflict resolution to a sample of managers. These
formats were paper-and-pencil, computerized page-turner, and video-based (multimedia). They
compared reactions to the multimedia version with pooled reactions to the other two formats and
found that managers perceived the multimedia version to have higher content and predictive
validity. Managers also found the multimedia version more enjoyable and were more satisfied
with the overall assessment process when completing that version of the test.
Meta-analytic findings show that White test takers score higher on average on SJTs than
do Black, Hispanic, and Asian examinees (Whetzel, McDaniel, & Nguyen, 2008). However,
video-based SJTs may exhibit lower adverse impact than their written counterparts (Lievens &
Sackett, 2006). Chan and Schmitt (1997) found that a video-based SJT demonstrated less adverse
impact against Black students than did a paper-and-pencil version of the test. Black and White
students’ scores on the written SJT were positively correlated with their scores on a test of
reading comprehension. Scores on the video-based SJT, however, were nearly unrelated to
reading comprehension scores, suggesting that the larger adverse impact seen on the written SJT
may have been due in part to its reading comprehension component. Jones and DeCotiis (1986)
make a related point about an advantage of video-based SJTs: individuals who may perform
poorly on written tests, due perhaps to English not being their native language, may perform
relatively better on video-based tests because they can see and hear the relevant information for
each item.

8

In research that makes use of both written and video-based versions of an SJT in studying
outcomes of interest, the typical approach is to begin with a video-based test and create a paperand-pencil version of it (see Lievens & Sackett, 2006 and Chan & Schmitt, 1997), rather than
vice versa. This makes sense given that researchers tend to acquire these expensive video-based
instruments from organizations rather than develop them on their own. Researchers creating
paper-and-pencil versions of existing video-based SJTs must make important decisions regarding
the amount of contextual information to transfer to the written test. As an example, Lievens and
Sackett (2006) developed a written SJT by simply transcribing the dialogue from the video-based
SJT. They excluded description of the non-verbal aspects of the video-based test from the written
version. This choice seems consistent with Weekley and Jones’s (1997) suggestion that videobased SJTs may introduce irrelevant contextual information and error into a test. There is little
consideration of the implications of such decisions in the research literature on SJTs.
By virtue of their medium, developers of video-based SJTs have to decide not only what
kind of contextual information to include but also who to cast as actors in the various scenarios
composing the test. Test developers might make the strategic choice to include actors of various
demographic backgrounds so as to signal to applicants that the hiring organization welcomes
diversity (Avery & McKay, 2006), but to the best of my knowledge, there is no research with
SJTs that examines what implications these casting choices have for either the test performance
or reactions of applicants from different demographic groups. Research has yet to show whether
carefully considering casting choices is imperative or of little import. For video-based SJTs that
test interpersonal skills and therefore present test takers with scenarios where two or more
individuals are experiencing an interpersonal dilemma, one might expect that responses test
takers give to these situations will be influenced by actors’ demographic characteristics to the

9

extent that individuals’ evaluations of and consequently behavior toward others is affected by
observed demographic characteristics. Theories and research findings in psychology, sociology,
and marketing can be brought to bear on this question.
Relational Demography
Relational demography is the term used to describe the comparison of the demographic
profiles of two or more individuals who interact with one another (Tsui, Egan, & O’Reilly,
1992). Research on relational demography in the workplace focuses on implications that the
demographic similarity between two or more individuals has for the quality of their interactions
with one another and for their individual work outcomes. Much of relational demography
research is based on the similarity-attraction paradigm (Byrne, 1971 as cited in Lau, Lam, &
Salamon, 2008), which proposes that individuals tend to be attracted to those who are similar to
them. Identifying with similar others reinforces one’s own identity (Steele, 1988) and has been
referred to as the self-continuity principle (e.g., Goldberg, Riordan, & Zhang, 2008).
The similarity-attraction paradigm considers the main source of interpersonal attraction to
be similarity in attitudes and research evidence indicates that attitude similarity does lead to
attraction (Huston & Levinger, 1978). Individuals can infer similarity in attitudes between
themselves and others by using demographic cues such as race or sex (Tsui et al., 1992). Indeed,
research has found race and sex similarity to be important factors in interpersonal attraction
(Ibarra, 1992; Thomas, 1990); individuals tend to perceive those demographically similar to
them more positively (Chatman & Flynn, 2001; Tsui & O’Reilly, 1989; Wesolowski &
Mossholder, 1997). Graves and Elsass (2005) cite research to suggest that individuals may have
more positive expectations about how similar others will behave and treat these individuals
accordingly.

10

Given its influence on interpersonal attraction, demographic similarity is expected to
result in a range of positive outcomes for interacting parties. Demographic dissimilarity, on the
other hand, is expected to result in more negative outcomes. Support is a frequent outcome of
interest. Goldberg, Riordan, and Zhang (2008) showed that White and Black employees may see
their supervisors as better role models when those supervisors are of the same race as them.
Similarly, employees may receive more instrumental support (i.e., support, such as challenging
assignments and sponsorship, that advances one’s career) from their assigned mentors when the
protégé and mentor are of the same race (Ensher & Murphy, 1997). Another study showed that
employees may perceive a higher level of family supportive supervision when their supervisors
are of the same race and/or sex as them relative to when supervisors are of a different race and
sex (Foley, Linnehan, Greenhaus, & Weer, 2006).
Performance ratings are another frequently considered outcome in this stream of research.
Tsui and O’Reilly (1989) found that supervisors may like their dissimilar subordinates less than
similar ones and give dissimilar subordinates lower performance ratings. The tendency for White
and Black superiors to give their dissimilar subordinates lower ratings has also been found in the
context of developmental performance ratings for managers (Mount et al., 1998). A metaanalysis of race differences in ratings of performance showed that White and Black raters tend to
give higher ratings to individuals of the same race as them (Kraiger & Ford, 1985). Similarly, in
the context of interview ratings, assessors have been shown to evaluate candidates of a different
race less favorably than candidates of their own race (Buckley et al., 2007). Additional evidence
of demographic similarity effects comes from cross-sectional survey research in the marketing
field, which considers the implications of demographic differences between communicators
featured in advertisements and viewers of these ads. This research indicates that Black and White

11

individuals tend to prefer ads that feature spokespeople of the same race as them (see Whittler,
1991 for a review). Findings of sex similarity effects are infrequently mentioned in the above
review because considerably less evidence of these effects has accrued relative to evidence of
race similarity effects. There has been investigation of moderators in research on sex similarity
effects, as well as in research on race similarity effects. Moderators help to elucidate conflicting
findings in relational demography research and are reviewed next.
Moderators of demographic similarity effects. While research has shown demographic
similarity to be related to positive outcomes in a variety of contexts, there has also been evidence
of asymmetrical effects of demographic similarity across demographic groups. Race as a
moderator is discussed first, and is followed by discussion of sex as a moderator. Finally, a
rationale for asymmetrical effects is provided.
Race similarity effects may be stronger for White than for minority (e.g., Black)
individuals. Goldberg (2005) found, for instance, that in the context of applicant selection, White
recruiters showed a stronger preference for applicants of their own race than did Black recruiters.
Reviews of rater-ratee race effects on performance ratings (Oppler et al., 1992; Pulakos et al.,
1989; Sackett & DuBois, 1991) following those of Kraiger and Ford (1985) cited earlier,
reported findings inconsistent with those of Kraiger and Ford for Black raters. While Whites
were again found to rate those of their own race higher, Blacks did not rate Blacks higher on the
vast majority of performance dimensions. In fact, Blacks may actually rate Whites slightly better
than they rate those of their own race (Sackett & DuBois, 1991). Finally, research into the effects
of race composition of work groups suggests that for Whites increasing difference from others in
the work group is associated with lower organizational attachment, while for nonwhites, being
different in race is not related to attachment (Tsui et al., 1992).

12

Evidence for asymmetrical sex similarity effects is more mixed and comes mainly from
investigations of work groups (as opposed to dyads). Tsui, Egan, and O’Reilly (1992) found that
being different from those in one’s group in sex has a more negative effect on men’s
psychological attachment to the group than on women’s. While increased differences in the sex
composition of the group were related to lower attachment for men, women responded to
increased differences in sex composition of the group with higher levels of attachment. Chatman
and O’Reilly’s (2004) findings were consistent with Tsui and colleagues’: men were more likely
to indicate a preference for leaving female dominated groups than were women to indicate a
preference for leaving male dominated groups. On the other hand, Kirchmeyer (1995), who
conducted a longitudinal study of the effects of demographic similarity to one’s workgroup on
outcomes (e.g., work group fit, supervisor support, organizational commitment, turnover, and
promotion), did not find sex to moderate the influence of demographic similarity on outcomes;
dissimilarity did not have a more negative influence on men than on women. In the case of
promotion as an outcome there was actually an unexpected finding in that men benefited from
being dissimilar (in sex) from the work group. Finally, Hinds and colleagues (2000) asked
students to designate individuals they would like to work with and did not find individuals to
show a preference for working with men. (In fact, they did not even find sex similarity effects;
sex similarity did not play a significant role in group member choice).
Asymmetrical effects of demographic similarity have been explained in terms of groups’
differences in social status. Chattopadhyay, Tluchowska, and George (2004) propose that
members of low status groups (e.g., females, minorities) may benefit less than members of high
status groups (e.g., males, Whites; Elsass & Graves, 1997) from identifying with other members
of their demographic groups because identifying with those of a low status does not enhance

13

one’s identity. Members of low status groups can enhance their identities by associating, instead,
with those in the out-group. Thus, unlike individuals of high status, who maintain a positive
identity by evaluating members of their out-group less favorably than members of their in-group
(Goldberg, 2005), those of low status may not view their out-group more negatively than their
in-group. It is important to note, however, that there may be individual differences in how highly
identified members of low status groups are with their in-group. Both the relational demography
and marketing literatures indicate that ethnic identification may moderate demographic similarity
effects. Experimental research studies indicate that preference for one’s own group is stronger
among those who are highly identified with their group (Whittler, 1989; Williams & Qualls,
1989).
Before wrapping up the discussion of moderators it is important to note that there is some
evidence in the marketing literature to show that the effects of demographic (dis)similarity for
high status group members are not always straight forward either. Attitudes toward other racial
groups appear to moderate the effects of race similarity on White individuals. Whittler and
DiMeo (1991) found that White individuals with high scores on a measure of prejudice reported
feeling stronger identification with and similarity (in terms of lifestyle, background, appearance,
dress, and values) to White actors than Black actors; Whites who received low scores on the
measure of prejudice identified equally with and saw themselves as equally similar to White and
Black actors.
Time frame for demographic similarity effects. The very short time frame over which
test takers view video-based SJT items might appear to pose a problem when trying to draw on
relational demography research to make predictions about the role demographic similarity will
play in respondents’ evaluations of (and consequently behavioral intentions toward) actors in an

14

SJT. On the contrary, demographic similarity effects will be especially relevant with the short
time. It is important to note that the effects of demographic (dis)similarity are expected to be
fairly immediate (and to occur even absent interaction between individuals; Tsui et al., 1992) and
to actually decrease with time (Watson, Kumar, & Michaelsen, 1993). The expected immediacy
of the effects is supported by models of impression formation, which suggest that individuals
process others’ social category (including demographic) information automatically (Brewer &
Harasty Feinstein, 1999; Fiske, Lin, & Neuberg, 1999). More specifically, research shows that
when one encounters a member of a certain group about whom he or she has no individuating
information—that is information specific to that individual as opposed to information about the
individual’s social category (inferred from characteristics such as race, sex, and age)—his or her
judgment of that group member will be influenced by that individual’s social category (Kunda &
Thagard, 1996 provide a discussion of this research). The effects of demographic (dis)similarity
should decrease with time, as Watson and colleagues (1993) suggest, because with time,
individuating information about individuals becomes available. Thus, one might expect the types
of effects that have been found in relational demography research to generalize to the context of
items on a video-based SJT, where respondents have little more than the actors’ demographic
information upon which to base their evaluations of these individuals.
Demographic similarity and behavioral intentions. Respondents’ evaluations
(affective, cognitive) of the actors in the videos, based on the demographic similarity between
the respondent and the target actor, can be expected to influence respondents’ behavioral
intentions. This expectation is consistent with research linking evaluations of targets with
behavior toward those targets. For example, teachers’ interactions with students who they
perceive to be more (less) capable tend to be characterized by better (worse) quantity and quality

15

(see Reyna, 2008 for a review of the consequences of stereotypes in classroom settings). In the
context of helping situations, evaluations of help seekers (how responsible are they for their
problems?) are often found to be related to helpers’ behavior and intentions (MacGeorge, 2003).
To summarize, the relational demography perspective, if applied to video-based SJTs,
would suggest that demographic (dis)similarities will have effects on respondents’ evaluations of
target actors. Evaluations of target actors, in turn, will influence respondents’ behavioral
intentions. Notably, particular types of situations may prove important in the operation of
demographic similarity effects. Given that individuals are socialized to behave in certain ways in
particular situations (Wilson, Lizzio, Whicker, Gallois, & Price, 2003), different situations may
elicit different patterns of behavior, and potentially temper relational demography effects on
evaluations and behavioral intentions. Two particular contexts that have received attention in the
social categorization literature are helping situations and conflict situations. These situations are
also commonly found in SJTs. In the next section helping situations will be reviewed and
hypotheses for helping-type SJT scenarios will be presented. Afterward, conflict situations will
be reviewed and hypotheses for conflict-type SJT scenarios will be proffered.
Helping Situations
Helping behavior is considered a form of citizenship behavior (―performance that
supports the social and psychological environment in which task performance takes place‖;
Organ, 1997, p. 95) and involves helping coworkers solve (or prevent) work-related problems
(Podsakoff, MacKenzie, Paine, & Bachrach, 2000). A meta-analysis conducted by Podsakoff,
Whiting, Podsakoff, and Blume (2009) found that organizational citizenship behaviors are
related to both individual-level (e.g., managerial ratings of employee performance, absenteeism,
turnover intent and turnover) and organizational-level (e.g., productivity, customer satisfaction,

16

unit-level turnover) outcomes. Earlier, Podsakoff and colleagues’ (2000) narrative review of the
literature tied helping behaviors, in particular, to indicators of organizational-level outcomes
(e.g., efficiency, performance quantity and quality, customer satisfaction). Given these
relationships between helping behavior (and organizational citizenship behavior in general) and
outcomes organizations value, Podsakoff and colleagues (2009) recommended that organizations
select individuals who will be likely to perform these behaviors. It is not surprising then, that
SJTs often include help seeking (or giving) situations to determine applicants’ willingness to
offer help to coworkers. Research on helping was consulted to determine the role that
demographic similarity may play in individuals’ propensity to provide help.
Helping and race similarity. Milliken and Martins (1996) observed that when
differences (e.g., racial, gender) between people can be clearly seen, biases, prejudices, and
stereotypes can cause individuals to respond to these differences in particular ways. Helping
researchers have been particularly interested in how individuals respond to observed differences
in race between themselves and those that they are in a position to help. Helping of Blacks
(relative to Whites), who have historically been the targets of negative attitudes and
discrimination in our society, is most often of interest in this research. Saucier, Miller, and
Doucet (2005) conducted a meta-analysis of studies looking at differences in White individuals’
helping of other Whites and Blacks. While they did not find evidence of universal discrimination
against Blacks, they did find that Whites helped Blacks less than they helped those of their own
race in the same situations when they had more ways to rationalize withholding help without
needing to implicate race. In other words, when the situation provided nonracist justification for
not helping a Black individual Whites helped Blacks less. Help given to Blacks decreased as
compared to help given to Whites as the following factors increased: time it took to help, risk for

17

the helper, difficulty of helping, and effort of helping. Findings were mostly consistent with the
theory of aversive racism (e.g., Hodson, Dovidio, & Gaertner, 2002), which proposes that while
consciously believing in equality between the races and in their own lack of racist attitudes,
Whites could experience discomfort (e.g., fear, nervousness) in the presence of Black
individuals. Whites are not expected to act on this discomfort and show hostility or obvious
discrimination because of the cognitive dissonance (i.e., lack of consistency between one’s
professed attitudes and one’s behavior) this would cause, but they may behave in ways that do
not appear obviously racist (e.g., avoid contact). Avoidance behavior will not appear to the nonhelper or outsiders as obviously racist when the motivation for it is ambiguous (i.e., can be
explained by other situational factors) (Saucier, Miller, & Doucet, 2005).
In a series of experiments, Kuntsman and Plant (2008) found that White participants
perceived an emergency situation to be less severe and felt less inclined to provide help when the
victim was Black as compared to when the emergency was identical but featured a White victim.
Drawing on the aversive racism perspective, the researchers explained Whites’ distorted
perceptions of the emergency when the victim was Black, in terms of White’s desire to both
avoid interracial contact with an individual of a race that elicits negative associations (e.g., Black
seen as bad) and avoid appearing biased against Blacks. Importantly, the researchers also found
that Black participants, unlike Whites participants, did not differ in their perceptions of the
emergency depending on the race (Black, White) of the victim, and were equally willing to help
the Black and White victims. This finding makes sense in light of research discussed earlier
which suggests that demographic similarity may not influence all groups of individuals in the
same way (Riordan & Shore, 1997).

18

Based on relational demography research and Saucier, Miller, and Doucet’s (2005) metaanalysis, one might expect that for situational judgment items with scenarios related to helping,
the level of help respondents provide will be influenced by whether the help-seeking actor in the
scenario (―non-focal actor‖) is of the same or a different race than the respondent. As research on
racism usually focuses on Blacks (Alvarez, Juang, & Liang, 2006), little has been done to
examine how race factors into help received by individuals of Asian backgrounds. Moreover, I
could find no research that has placed Asians into roles of helpers to examine their tendencies to
offer or withhold help to individuals of varying races. Therefore, findings from relational
demography research are used and research with White and Black individuals is extrapolated
from to make predictions about the effects of race mismatch between the respondent and the
non-focal actor for Asian in addition to White and Black individuals.
Extrapolating from Saucier and colleagues’ (2005) finding that White individuals helped
Blacks less when the situation provided non-race-related reasons for withholding help, I expect
that White respondents will offer less help to a minority non-focal actor than to a White nonfocal actor. This is likely to be the case because respondents will be able to use situational, raceunrelated cues in the test items to justify offering a low level of help to the minority non-focal
actor or withholding help altogether. Based on research showing asymmetrical effects of
demographic similarity for high and low status groups (probably because identifying with a
higher status group rather than with their own low status group is reinforcing for low status
individuals), one might expect that a racial mismatch between the respondent and the non-focal
actor will have a more negative influence on help offered by White as compared to minority
respondents.

19

Hypothesis 1: Respondent race will moderate the effects of the race mismatch between
the respondent and the non-focal actor on helping, such that the effects of race mismatch
will be more negative for White respondents viewing a minority non-focal actor than for

Helping

minority respondents viewing a White non-focal actor.

White Resp- White Resp- Black RespWhite Actor Minority Actor Black Actor

Black RespWhite Actor

Asian RespAsian Actor

Asian RespWhite Actor

Figure 1. Illustration of Hypothesis 1.

Helping and sex similarity. A considerable amount of research has investigated gender
effects on helping. In her review of the literature, Eagly (2009) notes that when it comes to
organizational citizenship behaviors, women (as compared to men) seem to perform more
behaviors (―altruism‖) that are targeted toward specific individuals who need assistance, while
men (as compared to women) appear to perform more behaviors (―civic virtue‖) that show extra
commitment to the organization (versus being aimed at particular coworkers). The former type of
citizenship behavior is of more relevance to the type of helping SJT items in the present
situation. I expect, therefore, that on average, female respondents will offer more help to nonfocal actors than will male respondents.

20

Hypothesis 2: There will be a main effect for respondent sex on helping, such that female
respondents will offer more help to non-focal actors than will male respondents.

Interestingly, meta-analytic evidence on the role of gender in helping (i.e., Eagly &
Crowley, 1986) does not support the notion of sex similarity effects in this context. Eagly and
Crowley (1986) concluded that men are significantly more likely to help women than they are to
help other men, and women are about as likely to help men as they are to help other women.
What this suggests then is interactive effects of target sex and respondent sex on helping. These
interactive effects may be related both to the consistency (or lack thereof) of targets’ behavior
with gender roles as well as men’s benevolent sexism toward women.
Behavior consistent with gender roles (commonly held beliefs about male and female
individuals; Eagly, 2009) tends to be met with approval whereas behavior at odds with gender
roles tends to be punished (e.g., Eagly, Karau, & Makhijani, 1995). Gender norms for men
prescribe autonomy and reliance on oneself (Auster & Ohm, 2000) and there is evidence to
suggest that when men are seen as departing from these proscriptions (e.g., by asking for help),
other men may react more negatively than women. In fact, women have been found to have more
egalitarian gender role orientations than men (e.g., Judge & Livingston, 2008). MacGeorge
(2003) found that men presented with a male’s request for help showed a greater tendency than
did women to blame the male (as evidenced by perceiving him as more responsible for his own
problems and becoming more angry at him). Women are stereotyped as being weak, needy,
whiny, and lower in competence relative to men (Kidder & McLean Parks, 2001). Thus, when
women ask for help, they are seen as acting consistently with expectations for their behavior.
Furthermore, paternalistic protectiveness (exhibited, perhaps, via politeness and courteous acts)

21

may be used to reward women whose behavior is appropriately guided by gender roles (Glick &
Fiske, 2001). Thus, when women ask for help, male helpers are expected to be more willing to
assist them due to their ―acting like women‖ than to assist males due to their being seen as
violating gender norms for males. Women, with their more egalitarian views toward gender
norms should provide equal assistance to male and female requesters.
Note that given the divergent evidence provided by research on helping and relational
demography in regards to gender and gender similarity effects, the relatively high
inconclusiveness of relational demography research in regard to these effects, and the higher
relevance of research on helping to the present study, more weight is being placed on findings
from helping research when forming the following hypothesis:

Hypothesis 3: Respondent sex will moderate the effects of non-focal actor sex on helping,
such that a female non-focal actor will receive more help than a male non-focal actor
when the respondent is male. Female and male non-focal actors will receive equal
amounts of help when the respondent is female.

Helping

Male Respondent
Female Respondent

Male Actor

Female Actor

Figure 2. Illustration of Hypothesis 3.
22

Race and sex considerations in helping. Because individuals belong to multiple groups
(race, gender), one needs to consider which of their identities individuals will use to relate to
actors in situational judgment items. Researchers (e.g., Whittler & DiMeo, 1991; Williams &
Qualls, 1989) often assume that individuals will primarily use race to identify with others,
regardless of whether or not identification based on other characteristics (e.g., gender) is possible
in the given situation. This is consistent with the ethnic prominence hypothesis proposed by
Levin and colleagues (2002), which suggests that in cases where multiple identities intersect,
ethnicity will be most salient. Brumbaugh (2009) questions this assumption and demonstrates
that identification with others is not always based mainly on race; it depends on contextual cues
which make race more or less prominent in a given situation. Indeed, Browne and Misra (2003)
in their review of the literature on the intersection of race and gender point out the
inconclusiveness in the findings (will ethnicity [Levin, Sinclair, Veniegas, & Taylor, 2002],
gender [Sidanius & Pratto, 1999], or both [Landrine, Klonoff, Alcaraz, Scott, & Wilkins, 1995;
Nelson & Probst, 2004] determine outcomes?) and argue for the need to carefully consider
context. Brumbaugh’s (2009) findings were actually not that surprising given that context has
been recognized earlier as influencing the salience of various identities (Ridgeway & SmithLovin, 1999 provide a review).
In the context of situations where an individual is asked for assistance, he or she would
be likely to consider the competence and work ethic of the individual seeking assistance.
However, when no additional information about the target beyond social category is available,
stereotypes (beliefs about social groups) will provide information about target individuals for
processing (Landy, 2008). When it comes to competence, women, as well as Blacks, are
stereotyped as being less competent than men and Whites, respectively. Accordingly, women
23

and Blacks are expected to need to ask for help. When it comes to work ethic, Blacks are
stereotyped as lazy while Asians are stereotyped as hard working (e.g., Reyna, 2008). Therefore,
Blacks might be seen as asking for help so as to avoid doing work themselves while Asians
might be seen as asking for help because they genuinely need assistance. Timberlake and Estes
(2007) demonstrated that stereotypes are amenable to categorization along dimensions of race
and sex salience (p.404). Using this scheme, the competence stereotype could be argued to have
high race and high sex salience while the work ethic stereotype could be argued to have high
race but low sex salience. To the extent that one or both of these stereotypes will be activated in
a particular helping situation, the sex, race, or both sex and race of the target will be salient to the
respondent.
Considering the two types of helping situations in the present study (these two situations
are described in the methods section), the competence stereotype is more likely to be relevant for
one situation while the work ethic stereotype is more likely to be relevant for the other. When the
competence stereotype is activated, and both race and sex is salient to respondents, the lowest
level of help is likely to be offered when a White male respondent is combined with a Black nonfocal actor as compared to other combinations of respondents with targets. This expectation is
based on the research already discussed. As explained earlier (recall Hypothesis 1), the effects of
race mismatch are expected to be more negative for White respondents viewing a Black nonfocal actor than for minority respondents viewing a White non-focal actor, and men are expected
to help other men less than they help women (recall Hypothesis 3).

24

Hypothesis 4: In a helping situation where the competence stereotype is activated, the
lowest level of help will be offered by White male respondents to male Black non-focal

White m. resp., m. White actor

White m. resp., m. Black actor

White m. resp., f. White actor

White m. resp., f. Black actor

White f. resp., m. White actor

White f. resp., m. Black actor

White f. resp., f. White actor

White f. resp., f. Black actor

Min. m. resp., m. White actor

Min. m. resp., m. Black actor

Min. m. resp., f. White actor

Min. m. resp., f. Black actor

Min. f. resp., m. White actor

Min. f. resp., m. Black actor

Min. f. resp., f. White actor

Min. f. resp., f. Black actor

Helping

actors as compared to other race and sex-based combinations of respondents with targets.

Figure 3. Illustration of Hypothesis 4.

When the work ethic stereotype is activated, and race is most salient to respondents, the lowest
level of help is likely to be offered when a White respondent is combined with a Black non-focal
actor. This expectation is, in effect, subsumed within hypothesis 1, but a separate hypothesis is
proposed because one particular type of helping situation is being considered at this time.

25

Hypothesis 5a: In a helping situation where the work ethic stereotype is activated, the
lowest level of help will be offered by White respondents to Black non-focal actors as

White respondent,
White actor

White respondent,
Black actor

White respondent,
Asian actor

Black respondent,
White actor

Black respondent,
Black actor

Black respondent,
Asian actor

Asian respondent,
White actor

Asian respondent,
Black actor

Asian respondent,
Asian actor

Helping

compared to other race-based combinations of respondents with targets.

Figure 4. Illustration of Hypothesis 5a.

Asians are excluded from being grouped with Blacks in the above hypothesis because of their
status as the ―model minority‖ (Alvarez et al., 2006). Recall the stereotype that Asians have high
work ethic. When this stereotype is activated, the level of help received by Asian non-focal
actors may even exceed the level of help received by White non-focal actors (the high status
group). (Notably, if this prediction is supported, there will be some inconsistency with
Hypothesis 1).

Hypothesis 5b: In a helping situation where the work ethic stereotype is activated, the
highest level of help will be offered to Asian non-focal actors.
26

Helping

Asian actor

Black actor

White actor

Figure 5. Illustration of Hypothesis 5b.

Conflict Situations
The term ―conflict‖ is used to describe a situation where interacting individuals have
apparently incompatible concerns (Thomas, Thomas, & Schaubhut, 2008). The success of
individual employees and the teams they work within depend in part on individuals’ ability to
deal effectively with conflict situations that arise (Tjosvold, 1998), with managers, for example,
spending 20 percent of their time, on average, dealing with conflict (Thomas, 1992, as cited in
De Dreu et al., 2001). As organizations become more diverse as well as more reliant upon teams
of individuals to work together effectively, of substantial concern is the potential for conflict
between individuals in demographically heterogeneous groups (King, Hebl, & Beal, 2009).
Given this concern, it is not surprising that organizations, as part of their selection process,
should wish to examine applicants’ (particularly for managerial positions) ability to effectively
handle conflict situations. Thus, interpersonally-oriented SJTs are likely to incorporate items
dealing with conflict situations (e.g., Chan & Schmitt, 2002; Mumford et al., 2008), or to be
27

composed entirely of conflict situations (e.g., Olson-Buchanan et al., 1998). Research on conflict
was consulted to determine the role that race and sex may play in how individuals choose to
handle conflict.
Several frameworks describing conflict situations have been proposed. One such
framework comes from K. W. Thomas’ work. Two dimensions are used for mapping the parties’
intentions: cooperativeness and assertiveness (Thomas et al., 2008). Cooperativeness refers to the
attempt to satisfy the other party; assertiveness refers to the attempt to satisfy one’s own
interests. Various combinations of these two dimensions (e.g., low cooperativeness with low
assertiveness, high cooperativeness with high assertiveness) constitute the type of conflict
resolution strategy an individual employs.
Conflict and race similarity. Not inconsistent with the relational demography
perspective, researchers generally expect individuals to exhibit a tendency towards cooperating
more with similar others than with dissimilar others (Tajfel, 1978, as cited by King et al., 2009).
Consistent with the earlier discussion of the time frame of demographic similarity effects,
cooperation is expected to be difficult at early stages of individuals’ interactions because based
on limited information about one another, individuals categorize each other (as being in the outgroup) according to visible demographic characteristics (Chatman & Flynn 2001). In support of
this expectation, Chatman and Flynn (2001) found that demographic diversity between
individuals within groups is negatively related to perceptions and development of cooperative
norms when the group is starting out. With regards to race differences between individuals in
particular, it is argued that one’s race will influence perceptions and subsequently reactions to
others during conflict situations (Davidson, 2001). Specifically, conflicts often start with an
episode of potentially provocative behavior by one of the parties, which the other party has to

28

interpret in order to decide how to respond. The characteristics of the parties, including race,
should help to determine what the interpretation will be. Based on ideas from relational
demography, one might expect less negative interpretations of provocative acts among racially
similar (i.e., in-group members) individuals (Davidson, 2001). Surprisingly, though, researchers
sometimes find main effects for target’s race and the nature of these main effects is not even
consistent. Sagar and Schofield (1980), for instance, found that children (both Black and White)
judged ambiguous aggressive behavior more negatively when it was done by a Black as opposed
to a White individual. In Davidson’s (2001) experiment, both Black and White participants
interpreted a White manager’s provocative act (taking credit for another individual’s work) more
negatively than a Black manager’s. These inconsistent findings may have something to do with
differences in situational context (e.g., Blacks are stereotyped as being more aggressive than
Whites, Whites are seen as more exploitative than Blacks), reaffirming again the need to
consider context in making predictions regarding race effects in various situations.
With regard to actual responses to conflict, research points to some potential differences
between racial groups. Davidson (2001) found that Whites prefer avoiding conflict more than
Blacks do (Blacks are more comfortable with the idea of being confrontational than are Whites),
but both groups endorse an avoidance strategy more so when considering conflict with those of a
different race (only Black and White were considered). According to the conflict framework
described above, an avoidance strategy is defined as being low on the assertiveness as well as
low on the cooperativeness dimensions (Thomas et al., 2008). The expectation that individuals
will show higher cooperativeness toward racially similar others is consistent with the relational
demography perspective. The expectation that they will show higher assertiveness toward
racially similar others may seem inconsistent with the relational demography perspective but

29

perhaps it is not. Just as one might argue that liking another individual more (due to similarity
with him or her) should make someone less assertive with that individual, one can also argue that
feeling more comfortable in an interaction with another individual (again, due to similarity with
him or her) should make someone more willing to be assertive (i.e., interpersonal comfort may
underlie willingness to be assertive). To the extent that findings from Davidson’s (2001) study
generalize, individuals may potentially show lower assertiveness and lower cooperativeness
when responding to conflict with an individual from a different racial group.

Hypothesis 6a: In responses to conflict situations there will be a main effect for
respondent-non-focal actor race match on assertiveness such that respondents will
endorse a more assertive response when the respondent and the non-focal actor are of the
same race than when they are of different races.

Assertiveness

2
Respondents
1

0
Race Match

Race Mismatch

Figure 6. Illustration of Hypothesis 6a.

Hypothesis 6b: In responses to conflict situations there will be a main effect for
respondent-non-focal actor race match on cooperativeness such that respondents will
30

endorse a more cooperative response when the respondent and the non-focal actor are of
the same race than when they are of different races.

Cooperativeness

2
Respondents
1

0
Race Match

Race Mismatch

Figure 7. Illustration of Hypothesis 6b.

Another consideration in conflict situations is whether or not a third party is brought in to
try to resolve the conflict. The decision to involve a third party may arguably be related to
cultural differences between individuals. Those who are collectivist (i.e., emphasis on the
interdependence of group members) may conceivably be more likely to seek the help of a third
party than those who are individualistic (i.e., emphasis on independence from group members
and self-reliance) because of the desire to avoid confrontation and possible damage to the
relationship with the other party (Giebels & Yang, 2009). While there is no complete consensus
on this being the case (see Oyserman, Coon, & Kemmelmeier, 2002), research suggesting that
Asians, Hispanics, and Blacks have more collectivist (defined in that research as orientation
toward the welfare of one’s larger community) tendencies than do Whites (Gaines et al., 1997)
has lead researchers to test for main effects of culture on preferences for help from third parties.
Davidson (2001) evaluated but did not find support for the prediction that Blacks would use
31

social support in conflict situations more so that would Whites. Similarly, Giebels and Yang
(2009) did not find differences in general preference for third party intervention between the
groups they classified as collectivist (Chinese individuals) and individualist (Caucasian
individuals). It may be the case, however, that collectivist groups’ preference for third party
intervention also depends on whether or not the other party is of the same racial background.
That is, collectivist individuals may prefer to get assistance from a third party in conflict
situations with members of their group more so than in conflict situations with members of an
out-group because maintaining relationships with those in the in-group is perceived as more
important. Some support for this position comes from research suggesting that Chinese
individuals will be more likely to adopt a competitive approach over a cooperative approach
unless they are dealing with in-group members with whom the importance of a long-term
relationship outweighs the importance of short-term gain (Koch & Koch, 2007 provide a critical
analysis of relevant research). Race match, then, might have a positive effect on collectivist
individuals’ use of third party intervention. There is little evidence to suggest, however, that race
match should have an influence on individualist individuals’ use of third party intervention.
(Note that this hypothesis will apply only when intervention has a reconciliatory goal. This point
is further elaborated on below.)

Hypothesis 7: There will be an interaction between race and respondent-non-focal actor
race match on preference for third party intervention when the goal is reconciliation.
Respondents belonging to groups classified as collectivist (Asians, Blacks) will show a
preference for third party intervention when paired with non-focal actors of the same race
as them as compared to when they are paired with non-focal actors of a different race.

32

Race match between respondents and non-focal actors will not influence preferences for

Preference for Third Party
Intervention

third party intervention among respondents classified as individualist (Whites).

2

Collectivist (Asian, Black)

1

Individualist (White)

0

Race Match

Race Mismatch

Figure 8. Illustration of Hypothesis 7.
The above hypothesis is likely to only hold true for one of the conflict situations in the
proposed study, not both. This is due to a divergence in the goals of third party intervention in
the two SJT scenarios. While in one (for which the above hypothesis is relevant) intervention
appears to have a reconciliatory goal, in the other intervention has more of a report-coworker’spoor-performance goal. The ideas that informed the above hypothesis (i.e., different groups’
differential desire to maintain good interpersonal relationships) logically lead to different
predictions depending on the goal of the third party intervention. When the coworker is of the
same race as them , members of collectivist groups should be more likely to prefer an
intervention when the goal is reconciliation, but less likely to prefer an intervention when the
goal is reporting the coworker’s poor performance. Reconciliation with a coworker helps to
maintain a relationship; getting a coworker into trouble is likely to harm the relationship. As for
the case when reconciliation is the goal of the intervention, there is little evidence to suggest that

33

race match should have an influence on individualist individuals’ use of third party intervention
when reporting a coworker’s poor performance is the goal. Thus, the following is expected for a
scenario where third party intervention has the goal of reporting a coworker’s poor performance:

Hypothesis 8: There will be an interaction between race and respondent-non-focal actor
race match on preference for third party intervention when the goal is reporting a
coworker’s poor performance. Respondents belonging to groups classified as collectivist
(Asians, Blacks) will show a preference for third party intervention when paired with
non-focal actors of a different race as them as compared to when they are paired with
non-focal actors of the same race. Race match between respondents and non-focal actors
will not influence preferences for third party intervention among respondents classified as

Preference for Third Party
Intervention

individualist (Whites).

2

Collectivist (Asian, Black)
1
Individualist (White)

0
Race Match

Race Mismatch

Figure 9. Illustration of Hypothesis 8.

Conflict and sex similarity. As previously noted, gender norms prescribe and proscribe
certain behaviors for men and women. Women are expected to be more nurturing and
34

subordinate, while men are expected to be more assertive and dominant (Graves & Elsass, 2005).
Accordingly, women are seen as more oriented than men toward maintaining relationships
(Wilson et al., 2003). Asserting one’s rights is perceived as detrimental to interpersonal
relationships and women who behave this way are seen as violating gender norms for women
(Wilson et al., 2003). Furthermore, research suggests there is a high level of agreement among
people regarding what is socially appropriate behavior for their own and opposite gender groups
in particular contexts (e.g., Wilson et al., 2003). For instance, Wilson and colleagues (2003)
found that when unfairly criticized, men agree that it is socially appropriate for them to be
proactive and direct in asserting that they have been evaluated unfairly. Women, on the other
hand, agree that it is fitting for them to be less direct and to allow the criticizer a chance to
explain his or her point of view. Women might compromise and accept some amount of
responsibility in a conflict situation so as to avoid damaging the relationship with the other party
(Wilson et al., 2003). This sort of behavior is, again, in line with gender role expectations for
women.
Given that women tend to endorse gender norms which dictate that women should use
more affiliative strategies in resolving interpersonal challenges and that men tend to endorse
norms that dictate that men are allowed to behave more assertively, female test takers are
expected to respond to conflict situations in a more cooperative and less assertive manner than
male test takers.

Hypothesis 9a: There will be a main effect for respondent gender on assertiveness when
responding to conflict, such that female respondents will respond to conflict with lower
assertiveness than will male respondents.

35

Hypothesis 9b: There will be a main effect for respondent gender on cooperativeness
when responding to conflict, such that female respondents will respond to conflict with
higher cooperativeness than will male respondents.

Behavior and evaluations of its effectiveness has been shown to be related to the relative
gender of the interacting parties (Carli, 1989). In an investigation of gender differences in social
influence within dyads, Carli (1989) found that both men and women agreed more and exhibited
more positive social behaviors (e.g., displays of positive affect, displays of solidarity) when
paired with a female interaction partner. Boys are reported to employ verbal, physical, and
indirect (e.g., shut out of group, ignore) aggressiveness less with girls than with other boys
(Russell & Owens, 1999). The same study also found girls to be more physically aggressive
toward boys than toward other girls. Both male and female adolescents use compromise as a
conflict resolution strategy more with females than with males and males display anger less with
females than they do with other males (girls, however, display equal amounts of anger with both
gender groups) (Shute & Charlton, 2006). Individuals are more accepting of ―masculine‖
behavior directed at men when discussing controversial topics (Carli, 1989) and perceive it as
appropriate to be more assertive in response to a man offering criticism than in response to a
woman offering criticism (Wilson et al., 2003, Experiment 2).Thus, a considerable amount of
research suggests that both men and women seem to perceive it more acceptable to behave in a
more masculine way toward men than toward women. These findings apparently reflect attitudes
of benevolent sexism toward women (Wilson & Gallois, 1985) and are not inconsistent with
findings from the research on helping reviewed earlier.

36

Given the research that suggests a general tendency to behave more positively toward
women (Carli, 1989) as well as the more specific tendency to be less assertive when responding
to a criticizing woman as compared to a criticizing man, asymmetrical effects of sex similarity
between respondents and non-focal actors in conflict-related SJT items are expected. Men are
expected to respond to male actors in a less affiliative manner (higher assertiveness, lower
cooperativeness) than to female actors while women are expected to respond to female actors in
a more affiliative manner (lower assertiveness, higher cooperativeness) than to male actors.

Hypothesis 10a: Respondent gender will moderate the effects of gender match between
the respondent and the non-focal actor on assertiveness when responding to conflict. For
male respondents, gender match will have a positive effect on assertiveness. For female
respondents, gender match will have a negative effect on assertiveness.

Assertiveness

2
Male Respondent
1

Female Respondent

0
Male Actor

Female Actor

Figure 10. Illustration of Hypothesis 10a.

Hypothesis 10b: Respondent gender will moderate the effects of gender match between
the respondent and the non-focal actor on cooperativeness when responding to conflict.

37

For male respondents, gender match will have a negative effect on cooperativeness. For
female respondents, gender match will have a positive effect on cooperativeness.

Cooperativeness

2

Male Respondent
1
Female Respondent

0
Male Actor

Female Actor

Figure 11. Illustration of Hypothesis 10b.

It is necessary to consider the role of gender in determining preferences for third party
interventions in conflict situations. Giebels and Yang (2009) did not find differences in general
preference for third party intervention between men and women. Interactive effects may be
possible, however. As discussed earlier, women are expected to be more nurturing than men
(Graves & Elsass, 2005) and more oriented toward maintaining relationships in general and
during conflict situations specifically (Wilson et al., 2003). Given evidence that women will tend
to endorse such gender norms (e.g., Wilson et al., 2003), a similar process may operate for
women wanting to preserve relationships in the face of conflict as was described earlier for
collectivist groups. Combined with the expectation that women will respond in a more affiliative
manner toward other women than toward men (recall Hypothesis 10), this yields the possibility
of a respondent gender by respondent-target gender match interaction on third party intervention
preferences. The nature of the interaction should again (recall the difference between hypotheses
38

7 and 8) be different depending on the type of conflict scenario (i.e., goal of reconciliation, goal
of reporting a coworker’s poor performance).
The effect of gender match on male respondents is somewhat harder to predict than the
effect of gender match for female respondents. Given that men are expected to respond to
women in a more affiliative manner than to other men (recall Hypothesis 10), like with female
respondents, their preferences may also be influenced by their gender (mis)match with the nonfocal actor. Alternatively, however, men may not want third party intervention regardless of the
gender of the non-focal actor because it may be seen as less than masculine to seek help in
resolving one’s interpersonal issues. I tentatively hypothesize that the first alternative will hold
true. Thus, it is expected that:

Hypothesis 11: When the goal of the conflict situation is reconciliation, there will be an
interaction between respondent gender and respondent-non-focal actor gender match on
preference for third party intervention of the following nature: a. Female respondents will
show a preference for third party intervention when paired with female non-focal actors
as compared to when they are paired with male non-focal actors; b. Male respondents
will show a preference for third party intervention when paired with female non-focal
actors as compared to when they are paired with male non-focal actors.

39

Preference for Third Party
Intervention

2

Female Respondent

1

Male Respondent

0
Gender Match

Gender Mismatch

Figure 12. Illustration of Hypothesis 11.
Hypothesis 12: When the goal of the conflict situation is reporting a coworker’s poor
performance, there will be an interaction between respondent gender and respondent-nonfocal actor gender match on preference for third party intervention of the following
nature: a. Female respondents will show less of a preference for third party intervention
when paired with female non-focal actors as compared to when they are paired with male
non-focal actors; b. Male respondents will show less of a preference for third party
intervention when paired with female non-focal actors as compared to when they are
paired with male non-focal actors.

40

Preference for Third Party
Intervention

2

Female Respondent

1

Male Respondent

0
Gender Match

Gender Mismatch

Figure 13. Illustration of Hypothesis 12.

Race and sex considerations in conflict. As discussed earlier, individuals belong to
multiple groups (race, gender), and it is necessary to consider if individuals will use race, gender,
or both to relate to actors in the conflict situational judgment items. It was also argued that
context needs to be considered as it can influence the salience of various identities (again,
Ridgeway & Smith-Lovin, 1999 provide a review). In the context of conflict situations, like in
the context of helping situations, certain stereotypes may be used to make judgments about the
non-focal actors in SJT items. When a conflict is based on a disagreement about the way work is
getting done, one might consider the competence of the other party. The competence stereotype,
it was argued earlier, has high race and high sex salience. When a conflict is based on
interpersonal friction, it is less clear what stereotypes may be used to make inferences about the
other party. However, it may be reasonable to expect that one might consider the agreeableness
and aggressiveness of the other party. Both agreeableness-related and aggressiveness-related
stereotypes arguably have high race and high sex salience. Given that Asians are seen as valuing
relational harmony over personal interests (Brett & Kopelman, 2004, as cited by Liu, 2009), they

41

may be stereotyped as being more agreeable than other groups. Women are stereotyped as
valuing relational harmony more so than men (Wilson et al., 2003) and therefore might be
expected to be more agreeable than men. When it comes to aggressiveness, Blacks are
stereotyped as being more aggressive than other groups (e.g., Harrison & Esqueda, 2001), and
women are expected to be less aggressive than men (Huddy & Terkildsen, 1993).
Given that both types of conflict situations in the present study (disagreement about the
way work is getting done, interpersonal friction) are likely to elicit stereotypes with high race
and high sex salience, both race and sex of the target should be salient to the respondent.
Predictions are therefore made with both expected race and expected sex effects in mind. First,
considering assertiveness, the lowest level of assertiveness in responses is likely to be exhibited
when females respond to non-focal actors of the same gender but of a different race, as compared
to other sex and race-based combinations of respondents with targets. This expectation is based
on the research already discussed. As proposed earlier, female respondents should respond to
conflict with lower assertiveness than male respondents (Hypothesis 9a), and respondents are
expected to be less assertive with non-focal actors of a different race (Hypothesis 6a). The
highest level of assertiveness is likely to be exhibited by male respondents paired with non-focal
actors of the same gender and race. This prediction is based on the expectations that men will be
more assertive than women (Hypothesis 9a), that men will respond to male non-focal actors
more assertively than to female non-focal actors (Hypothesis 10a), that more assertive responses
will be seen in responses to those of the same race (Hypotheses 6a).

42

Hypothesis 13a: In conflict situations the lowest level of assertiveness will be exhibited
by female respondents paired with a female non-focal actor of a different race as
compared to other sex and race-based combinations of respondents with targets.

Assertiveness

2

1

Female respondent,
male actor of diff
race

Female respondent,
male actor of same
race

Female respondent,
female actor of diff
race

Female respondent,
female actor of same
race

Male respondent,
female actor of diff
race

Male respondent,
female actor of same
race

Male respondent,
male actor of diff
race

Male respondent,
male actor of same
race

0

Figure 14. Illustration of Hypothesis 13a.

Hypothesis 13b: In conflict situations the highest level of assertiveness will be exhibited
by male respondents paired with a male non-focal actor of the same race as compared to
other sex and race-based combinations of respondents with targets.

43

Assertiveness

2

1

Female
respondent, male
actor of diff race

Female
respondent, male
actor of same race

Female
respondent, female
actor of diff race

Female
respondent, female
actor of same race

Male respondent,
female actor of
diff race

Male respondent,
female actor of
same race

Male respondent,
male actor of diff
race

Male respondent,
male actor of same
race

0

Figure 15. Illustration of Hypothesis 13b.

Next, considering cooperativeness, the lowest level of cooperativeness should be seen
when male respondents are paired with non-focal actors of the same gender but a different race
as compared to other sex and race-based combinations of respondents with targets. This
expectation is based on research discussed earlier. It was proposed earlier that men will show
lower cooperativeness than women (Hypothesis 9b), and that respondents will be less
cooperative with those of a different race (Hypothesis 6b). The highest level of cooperativeness
should be seen when female respondents are paired with non-focal actors of the same gender and
race as compared to other sex and race-based respondent-target combinations. This expectation
is also based on propositions put forth earlier. Specifically, women are expected to be more
cooperative than men (Hypothesis 9b) and to respond to female non-focal actors with higher
cooperativeness than to males (Hypothesis 10b). Respondents are expected to show more
cooperativeness toward those of the same race as them (Hypothesis 6b).

44

Hypothesis 14a: In conflict situations the lowest level of cooperativeness will be
exhibited by male respondents paired with male non-focal actors of a different race as
compared to other sex and race-based combinations of respondents with targets.

Cooperativeness

2

1

Female respondent,
male actor of diff
race

Female respondent,
male actor of same
race

Female respondent,
female actor of diff
race

Female respondent,
female actor of
same race

Male respondent,
female actor of diff
race

Male respondent,
female actor of
same race

Male respondent,
male actor of diff
race

Male respondent,
male actor of same
race

0

Figure 16. Illustration of Hypothesis 14a.

Hypothesis 14b: In conflict situations the highest level of cooperativeness will be
exhibited by female respondents paired with a female non-focal actor of the same race as
compared to other sex and race-based combinations of respondents with targets.

45

Cooperativeness

2

1

Female respondent,
male actor of diff
race

Female respondent,
male actor of same
race

Female respondent,
female actor of diff
race

Female respondent,
female actor of
same race

Male respondent,
female actor of diff
race

Male respondent,
female actor of
same race

Male respondent,
male actor of diff
race

Male respondent,
male actor of same
race

0

Figure 17. Illustration of Hypothesis 14b.

Finally, considering preference for third party intervention in the conflict situation,
different expectations apply for the two types of situations considered (interpersonal friction,
where the goal of third party involvement would be reconciliation; disagreement about the way
work is getting done, where the goal of third party involvement would be reporting a coworker’s
poor performance). When the goal is reconciliation, the highest preference for third party
intervention should be exhibited when female respondents from groups considered collectivist
are paired with non-focal actors of the same gender and race. This hypothesis is based on the
expectation that respondents from collectivist groups will show more of a preference for third
party intervention when paired with non-focal actors of the same race (Hypothesis 7) and that
respondents will show more preference for third party intervention when paired with female nonfocal actors (Hypothesis 11).

Hypothesis 15: In a conflict situation where the goal of involving a third party is
reconciliation, the highest preference for third party intervention will be exhibited by
46

collectivist (Asian, Black) respondents paired with a female non-focal actor of the same

2

1

White respondent,
female actor of diff
race

White respondent,
female actor of same
race

White respondent,
male actor of diff race

White respondent,
male actor of same
race

Minority respondent,
female actor of diff
race

Minority respondent,
female actor of same
race

Minority respondent,
male actor of diff race

0
Minority respondent,
male actor of same
race

Preference for Third Party Intervenion

race.

Figure 18. Illustration of Hypothesis 15.
When the goal is reporting a coworker’s poor performance, the highest preference for
third party intervention should be exhibited when male respondents from groups considered
collectivist are paired with non-focal actors of a different race. This hypothesis is based on the
expectation that respondents from collectivist groups will show more of a preference for third
party intervention when paired with non-focal actors of a different race (Hypothesis 8) and that
respondents will show more of a preference for an intervention when the non-focal actor is male
(Hypothesis 12).

47

Hypothesis 16: In a conflict situation where the goal of involving a third party is
reporting a coworker’s poor performance, the highest preference for third party
intervention will be exhibited by collectivist (Asian, Black) respondents paired with a

2

1

White respondent,
female actor of diff
race

White respondent,
female actor of same
race

White respondent,
male actor of diff race

White respondent,
male actor of same
race

Minority respondent,
female actor of diff
race

Minority respondent,
female actor of same
race

Minority respondent,
male actor of diff race

0
Minority respondent,
male actor of same
race

Preference for Third Party Intervenion

male non-focal actor of a different race.

Figure 19. Illustration of Hypothesis 16.

Applicant Reactions
Applicant reactions to selection procedures are an important area of research and one
with considerable implications for organizations. Organizational attractiveness, and intentions to
accept a job offer, recommend the company to other people, undertake legal challenges, and buy
company products have all been tied to applicant reactions (Hausknecht, Day, & Thomas, 2004).
Meta-analytic findings show no significant main effects of race or sex on applicant reactions
(e.g., perceptions of the fairness of the process and the outcomes received) (Hausknecht et al.,

48

2004). In the related research area on reactions to organizational recruitment there is evidence to
show that the race and sex of individuals included in recruitment campaigns as representatives of
the types of people in the hiring organization relative to the race and sex of the applicant does
have an effect on applicant reactions. Researchers in the area of targeted recruiting have devoted
substantial attention to examining the determinants of minority and female applicants’ reactions
to recruitment tactics (see Avery & McKay, 2006 for a review). Research has found that
minorities (Blacks, Hispanics) react positively to racial diversity in advertisements while Whites
are little influenced by level of diversity in ads (Avery, Hernandez, & Hebl, 2004; Perkins,
Thomas, &Taylor, 2000). While less research has been done to evaluate Asians’ reactions to
diversity, they have been shown to react more positively toward ads featuring Asians that toward
ads featuring Whites (Martin, Lee, & Yang, 2004). Notably, reactions toward diversity are likely
to be influenced by attitudes viewers of ads hold (e.g., level of racial identification, other-group
orientation, level of prejudice) (Avery, 2003; Whittler, 1991).
Interestingly, minorities (Blacks, Hispanics) seem to judge diversity based on minority
group representation, as opposed to simply own group representation, reacting positively as long
as either Blacks or Hispanics are well represented (Avery et al., 2004). These findings may
arguably be extended to predict that these groups would also react positively to the presence of
Asians, another minority group, and that Asians would react positively toward representation of
these other groups in ads. White respondents’ reactions, in the meantime, should not be affected
by the race of the non-focal actors.

49

Hypothesis 17: Minority (Black, Asian) respondents will react more positively when the
SJT features Black or Asian non-focal actors than when the SJT features White non-focal
actors.

Like racial minorities are in relation to Whites, women are more attuned than men to
diversity information (Avery & McKay, 2006). As explained later, the focal actor in the videos
will always be female to try to simplify somewhat the design of the study. Given that the focal
actor will always be female, in cases where the non-focal actor is also female, there will actually
be no gender diversity (because of the absence of men). Therefore, women will probably react
more positively to the SJT when the non-focal actors are male.

Hypothesis 18: Female respondents will react more positively when the non-focal actors
are male.

In summary, existing research in the areas of relational demography, helping, and conflict
suggests that different groups of test takers may respond to video-based situational judgment
items differently depending on the race and gender of the actors in the videos. Research on
applicant reactions to diversity in organizational recruitment materials suggests that test taker
reactions to the overall test may likewise differ as a function of actor demographics. Next, I
describe the method that was used to test the helping-, conflict-, and reactions-related hypotheses
reviewed above.

50

METHOD
Design
The video-based SJT included two types of scenarios (help seeking, conflict) with two
scenarios per scenario type. Thus, participants saw four scenarios, the order of which was
counterbalanced.
The demographic characteristics of the focal actor (whose role the respondent was asked
to assume) in each scenario were kept constant across study conditions. The focal actor was
always a young White female. This was done in the interests of keeping the study design
manageable. Though there is no guidance on this in the literature of which I am aware, the
expectation is that the race and gender of the focal actor had little influence on the ability of the
respondent to assume that individual’s role when responding to SJT items. That is, there is little
reason to believe that a respondent would actually try to ―get inside the head‖ of the person
whose role he or she needed to assume when determining what course of action to take as
opposed to simply putting him or herself in the focal actor’s shoes. It is more reasonable to
expect that the respondent approached the situation from his or her own perspective when
determining the best response as that is what respondents were explicitly instructed to do. If the
respondents did in fact approach each situation from their own perspective, the race and gender
of the focal actor should have mattered little. However, each focal actor was White and female,
given that most study participants were expected to be White females (because of the
composition of the psychology participant pool).
Across study conditions, the White female playing the focal person in a given SJT item
was always the same actress. The non-focal actor, however, was varied to create six different
versions of each of the four scenarios. That is, any given respondent saw four scenarios in all of
which the non-focal actor was of one race and gender. Yet, across the four scenarios that this
51

respondent saw, the non-focal actors were played by different actors. To illustrate, a respondent
might have seen 4 scenarios in which he/she always assumed the role of a White female to
respond to a coworker who was always an Asian male, but across the 4 videos, all the actors
were different Asian males.
The study assumed a 3x2x2x2 design (race of respondent: White, Black, Asian; race
match between respondent and non-focal actor: yes, no; sex of respondent: male, female; sex
match between respondent and non-focal actor: yes, no). The age of the non-focal person (as
well as the focal person’s, actually) was kept approximately constant, also in the interests of
making the design somewhat more manageable.
Sample
The sample for this study was drawn from the psychology participant pool at Michigan
State. The sample consisted of 374 students, of which 51% were female. The racial breakdown
was as follows: 72.5% White, 11% Asian, 9.4% African American/Black, .3% Native Hawaiian
or other Pacific Islander, 4.3% multi-racial, and 2.7% ―other‖. Average age was 19.97 (SD =
2.68).
Procedure
Students came into a computer lab to participate in the study. A research assistant briefly
instructed them on what they would be doing and handed out consent forms for them to sign.
After reading and signing their consent forms students proceed to take the four-item SJT test.
The test was preceded with several paragraphs asking students to imagine themselves as job
applicants taking an interpersonal test as an initial step in the process of applying for a technician
position at an attractive organization. This description was adapted from Bauer and colleagues
(2001) who used it as a manipulation of organizational attractiveness with a college student

52

sample, and modified to fit the present study. The version used in this study is provided in
Appendix A. Along with this manipulation respondents were also presented with a job
description for an entry level plant technician.
Participants viewed the four SJT items on a computer screen, in medialab. Each video
was followed by a screen where students indicated their likelihood of responding to the situation
just viewed in each of five or six different ways, and what they were most and least likely to do.
After completing the SJT students answered questions about their reactions. The order of the
reactions measures (opportunity to perform, perceived job relatedness, attraction, compatibility)
was counterbalanced such that half of the participants completed the opportunity to perform and
perceived job relatedness measures first and the other half completed the attraction and
compatibility measures first. This was done in case considering the fairness of the test should
influence evaluations of the hypothetical organization or vice versa. Following the reactions
measures, participants answered questions about their demographic backgrounds, rated the
attractiveness of the actors, and indicated these actors’ race and gender. Upon completion,
students received a sheet with information about situational judgment tests (the required
educational portion of the study).
Measures
Situational Judgment Test (SJT) Scenarios. Four scenarios were chosen from a full
video-based interpersonal abilities SJT used in the past by a large, global corporation as part of
its selection process for plant technicians. The test was developed to represent typical work
situations encountered by plant technicians. Incumbent technicians provided critical incidents
along with effective and ineffective responses to each. A concurrent validation design was used
to establish the effectiveness of the test for predicting plant technician performance. The test was

53

validated against performance ratings for approximately 400 technicians. Ratings were provided
by technicians’ immediate supervisors.
All four SJT scenarios selected for this study dealt with interpersonal situations occurring
between coworkers in an organization. More specifically, two scenarios pertained to help seeking
and two pertained to interpersonal conflict. The scripts for each of these four scenarios and the
accompanying questions are presented in Appendix C. Notes are included identifying which
stereotypes are pertinent to each scenario as well as what the goal of intervention would be for
those situations that pertain to conflict.
The scripts of these four scenarios were used to film six different versions of each of
these scenarios. The twenty four scenario clips were filmed with the assistance of WKAR,
Michigan State’s on-campus TV station. Graduate and undergraduate students were recruited to
act in these clips and were rewarded with $20 gift cards ($40 for focal actors). Once the clips
were filmed, medialab was used to organize them into a form that was shown to study
participants. Each scenario was followed by a screen where the respondent was asked to indicate
the likelihood of his/her responding to the situation in each of five or six different ways.
Likelihood ratings were provided for every option on a scale anchored by (1) not at all likely and
(5) very likely. Respondents also chose one option they would most likely do and one option they
would least likely do (e.g., ―If you were Jane, what would you MOST likely do?‖ and ―If you
were Jane, what would you LEAST likely do?‖)
SJT scoring. There are three approaches to scoring and examining responses to the SJT
in this study. The first, and the one most suited for testing the hypotheses in this study, is by
weighting likelihood ratings test takers gave to each response option according to the level of
helping, assertiveness, cooperation, or third party intervention a given response option

54

represents. Weighted likelihood ratings for the set of response options can then be summed up to
form scores representing helping, assertiveness, cooperativeness, and third party intervention.
Each test taker would then have scores on each of these dimensions: 2 helping scores for the two
helping SJT items and 2 assertiveness, 2 cooperativeness, and 2 third party intervention scores
for the two conflict SJT items.
To determine the level of relevant construct(s) represented by each SJT response option
(i.e., the item weights), judges were asked to provide ratings of options. They were recruited on
an individual basis through personal appeals and were unaware of the study’s hypotheses. The
final sample consisted of 40 judges. They were graduate students (51.3%), college graduates
(25.6%), college students (20.5%), and doctorate degree holders (2.6%). One judge did not
indicate level of education. They were almost evenly split in terms of gender (52.5% male).
Judges were White (72.5%), Asian (15%), Black (2.5%), and other/multi-racial (10%).
These individuals rated response options to the two conflict items on level of
assertiveness [using a scale from (1) very unassertive to (5) very assertive], cooperativeness
[using a scale from (1) very uncooperative to (5) very cooperative], and third party intervention
[using a scale from (1) none to (5) very high amount]. They rated response options to the two
helping items on level of helping [using a scale from (1) very unhelpful to (5) very helpful].
A second way to score, or in this case simply examine, responses to the SJT is to, again,
use the likelihood ratings test takers assigned to response options, but not weight these by the
corresponding level of a construct of interest (e.g., assertiveness). When this approach is used,
any analyses of responses are done at the level of the individual response options. If differences
between [groups of] test takers emerge, one has to take a close look at the individual response
options to come up with a substantive explanation of why likelihood ratings on these response

55

options should differ between [groups of] test takers. This is a different type of approach to
explanation than scoring method 1 (which allows one to look across response options in analyses
and speak about the dimensions on which responses of different [groups of] test takers may
differ). As such, method 2 is somewhat less amenable to testing the dimension-based hypotheses
in this study.
A third way to score the SJT is according to response effectiveness as determined by the
test developer. Respondents’ most and least likely responses to the four SJT items were scored
according to the scoring program used by the organization that provided their SJT for this study.
A most likely (ML) response was assigned a score of 1 if it matched what the company
considered to be a desirable response to the situation, a score of -1 if it matched a response
considered an undesirable option, and a score of 0 if it was a match for neither a desirable nor an
undesirable response. A least likely (LL) response was assigned a score of 1 if it matched what
the company considered to be an undesirable response to the situation, a score of -1 if it matched
a desirable response, and a score of 0 if it was a match for neither an undesirable nor a desirable
response. To clarify what defined ―desirable‖ and ―undesirable‖ it should be noted that the test
developer based the scoring key on response frequencies obtained from the sample of technicians
used to validate the test. Responses chosen most often by incumbents (specifically, by 100 or
more respondents) were considered desirable responses and received the highest scores.
Responses chosen least often by incumbents (specifically, by 0-25 respondents) were considered
undesirable responses and received the lowest scores. The zero scores were assigned to response
options chosen by 26 to 99 technicians in the sample. The resultant scoring key was reviewed by
personnel at the company to make sure that the scoring of all the response options was consistent
with performance expectations for plant technicians.

56

In this study the scores for the most likely response options were added up across the 4
SJT items to come up with an ML score, and scores for the least likely response options were
added up across the 4 SJT items to come up with an LL score. The ML and LL scores were then
combined to get a total SJT test score for each study participant. This approach to scoring the
SJT is most appropriate for examining performance on the SJT and identifying any differences
between groups of test takers that may exist. Similar to method 2 described above, the most and
least likely responses test takers provided can be examined for differences in which response
options were chosen most frequently by different groups. Thus, frequencies would be the
dependent variables in analyses (versus dimensions in method 1 and likelihood ratings in method
2). Like with method 2 one would then look closely at the individual response options to suggest
a substantive explanation for any differences. Another possibility is to separately use test takers’
scores on the 4 scenarios as dependent variables; helping item scores could be used in examining
hypotheses pertaining to helping and conflict item scores could be used in examining hypotheses
pertaining to assertiveness, cooperativeness, and intervention.
All three methods of scoring and examining SJTs were applied to various extents when
analyzing the results of this study.
Applicant reactions. Reactions to the SJT were operationalized in terms of perceptions
of opportunity to perform, perceptions of job relatedness, and attraction to and compatibility with
an organization administering such an SJT. Perceptions of opportunity to perform were measured
with four items adapted from Bauer and her colleagues (2001). Answers were provided on a five
point scale ranging from (1) strongly disagree to (5) strongly agree. A sample item is ―This test
allowed me to show what my interpersonal skills are.‖ Reliability was α = .84 for the measure.
The items are provided in Appendix D.

57

Perceived job relatedness was measured with four items adapted from Bauer and
colleagues (2001). Answers were provided on a five point scale ranging from (1) strongly
disagree to (5) strongly agree. A sample item is ―Doing well on this test means a person can do
the technician job well.‖ Bauer and colleagues found two of their job relatedness items loaded on
one factor (―job relatedness—predictive‖) and two loaded on a second factor (―job relatedness—
content‖). Reliability was α = .81 for the predictive items and α = .83 for the content items. The
items are provided in Appendix E.
Respondents were asked to indicate their attraction to the organization to which they
were hypothetically applying. Attraction was assessed with six items adapted from Perkins and
colleagues (2000). A sample item is ―I would speak to a company representative about the
possibility of employment.‖ Individuals indicated their agreement with each item using a scale
ranging from (1) strongly disagree to (5) strongly agrees. Reliability was α = .88 for this scale.
The items are provided in Appendix F.
Respondents were also asked to indicate their perceived compatibility with the
hypothetical organization. Compatibility was assessed with four items adapted from Perkins and
colleagues (2000) including ―I would feel at home working for an organization like this.‖
Individuals indicated their agreement with each item using a scale ranging from (1) strongly
disagree to (5) strongly agrees. Reliability was α = .85 for this scale. The items are provided in
Appendix G.
Demographic questions. Respondents answered questions about their age, gender,
ethnicity, race, year in college, GPA, and intended major. The list of demographic questions is
provided in Appendix H.

58

Coworker attractiveness. Because physically attractive individuals are initially liked
more than less attractive individuals (Berscheid & Walster, 1974), the attractiveness of
coworkers (i.e., non-focal actors) was measured so that it could be controlled for in the analyses.
Respondents were presented with photos of all the actors that appeared in the videos they
watched and asked to rate the attractiveness of each on a scale of (1) very unattractive to (5) very
attractive.
Manipulation checks. To make sure that the gender and race manipulations worked,
respondents were presented with photos of all the actors that appeared in the videos they watched
and asked to indicate the race and gender of each. The response options were Asian female,
Asian male, African-American/Black female, African-American/Black male, Middle Eastern
female, Middle Eastern male, White female, White male, and cannot tell. Even though none of
the actors were Middle Eastern, this was included as a choice to make these questions seem more
difficult (with the goal of discouraging respondents from picking unreasonable options just to
spite the experimenter for asking for very obvious responses).
Familiarity with actors. At the end of the experiment, respondents were asked whether
they were acquainted with any of the actors they saw in the four videos they watched (yes, no).
Those who responded ―yes‖ were presented with numbered photos of the individuals they saw in
the videos and asked to indicate the number(s) of the individual(s) with whom they were
acquainted.
Next, the results of the study are reviewed. Tests of the hypotheses presented earlier as
well as exploratory analyses are described.

59

RESULTS
Manipulation Checks
Twenty nine respondents of the 374 failed to identify the race and/or gender of the nonfocal actor in conflict situation 1, 28 of conflict situation 2, 12 of helping situation 1, and 27 of
helping situation 2. It should be noted that I did not consider identifying a Caucasian non-focal
actor as Middle Eastern to be a failure of the manipulation check because Middle Easterners are
technically Caucasian. Individuals who failed the manipulation check for a particular situation
were not included in the analysis of that situation (e.g., respondents who failed the manipulation
check for help seeking situation 2 were not included in analyses of help seeking situation 2).
Analyses were done to try to understand why so many respondents failed one or more of
the manipulation checks. As a group, those who failed at least one manipulation check were not
2

different from those who passed all the manipulation checks in terms of race [ (5, N = 374) =
2

2.37, ns], gender [ (1, N = 374) = 1.39, ns], or age [t(371) = .08, ns]. Though there were no
demographic differences between the groups, there were differences in response latencies on the
manipulation checks. Respondents who failed the manipulation check for conflict situation 1
spent significantly more time (recorded in milliseconds) on the question asking them to indicate
the demographic profile of the non-focal actor (i.e., the manipulation check) (M = 9993.24 ms,
SD = 5852.47) than those who passed the manipulation check for that situation (M = 5608.03 ms,
SD = 4031.25), t(30.27) = 3.96, p < .001. Respondents who failed the manipulation check for
conflict situation 2 spent significantly more time on the question (M = 6008.18 ms, SD =
2847.97) than those who passed the manipulation check for that situation (M = 2948.95 ms, SD =
2093.55), t(372) = 7.22, p < .001. Respondents who failed the manipulation check for helping
situation 1 spent marginally more time on the question (M = 4567.92 ms, SD = 2629.83) than
60

those who passed the manipulation check for that situation (M = 2894.47 ms, SD = 1953.75),
t(11.41) = 2.18, p = .051. Finally, respondents who failed the manipulation check for helping
situation 2 spent significantly more time on the question (M = 4455.07 ms, SD = 2434.11) than
those who passed the manipulation check for that situation (M = 2219.59 ms, SD = 1327.90),
t(27.22) = 4.72, p < .001. This pattern of differences suggests that respondents who failed the
manipulation check(s) were not unduly quick in identifying the actors (as opposed to tending to
simply rush through the questions and making errors because of that).
The actual responses of those who failed the manipulation check(s) were examined to
determine the nature of the identification difficulties their longer response times suggest. Table 1
provides a summary of the nature of respondents’ demographic profile identification difficulties
across the four manipulation checks. Overall, respondents had the most trouble correctly
identifying the demographic profile of the Black male and the White non-focal actors. Across
conditions the Middle Eastern response option was the biggest distracter for those who did not
correctly identify the demographic profiles of non-focal actors. The fact that some respondents
seemed to have misidentified actors’ gender (due to inattentiveness most likely, as opposed to
genuine identification difficulties, as none of the actors’ sex was ambiguous) also contributed to
manipulation check failures. Tables in Appendix I show the breakdown of incorrect responses to
the manipulation checks for the 4 SJT items individually. As a reminder, individuals who failed
the manipulation check for a particular situation were not included in the analysis of that
situation.

61

Table 1. Distribution of incorrect responses across manipulation checks
East
East
Asian
Asian
Black
Black
Caucasian
Condition
Female
Male Female
Male
Female
X
7.1%
---Asian Female non-focal
X
---16.7%
Asian Male non-focal
--X
-33.3%
Black Female non-focal
-4.3%
X
-4.3%
Black Male non-focal
8.3%
---X
White Female non-focal
-5%
--30%
White Male non-focal
Note. Bolded entries indicate errors based solely on gender error.

62

Caucasian
Male
----4.2%
X

Middle
Eastern
Female
64.3%
-33.3%
----

Middle
Eastern
Male
-50%
-56.5%
---

Cannot
tell
28.6%
33.3%
33.3%
34.8%
87.5%
65%

Total
N
14
12
3
23
24
20

Familiarity with Actors
Thirty four respondents indicated knowing at least one of the actors in the videos.
Specifically, for conflict situation 1, 7 knew the focal actor and 10 knew the non-focal actor. For
conflict situation 2, 9 knew the focal actor and 5 knew the non-focal actor. For helping situation
1, 7 knew the focal actor and 3 knew the non-focal actor. For helping situation 2, 2 knew the
focal actor and 13 knew the non-focal actor. As discussed later, familiarity with actors did not
have much influence on test takers’ responses.
Item Weights
In order to aggregate judges’ perceptions of what level of a particular construct (e.g.,
cooperativeness, assertiveness, third party intervention, helping) each SJT response option
represents, and using means to represent collective perceptions of response options, an adequate
level of inter-rater agreement (indicator of the extent to which individuals provided the same
ratings) had to be demonstrated (Kozlowski & Hattrup, 1992). James et al.’s (1984) rWG index
for assessing agreement among a set of judges rating one target using a single item was used to
assess agreement for each of the 44 items (ratings of one response option along one dimension
represent a single item). Before estimating rWG values, ratings that were more than three
standard deviations away from the mean of any given item were judged to be outliers and
removed. Sixteen outliers were removed from the data. A uniform null distribution was used to
estimate rWG values. Due to initially low estimates of rWG, the decision was made to collapse
the 5 point scale used by judges into a three point scale. For cooperativeness, assertiveness, and
helping ratings, ratings of 1 and 2 were collapsed into 1’s (low), ratings of 3 were recoded into
2’s (neutral), and ratings of 4 and 5 were collapsed into 3’s (high). For third party intervention

63

ratings, ratings of 1 remained as 1’s (none), ratings of 2 and 3 were collapsed into 2’s (medium)
and ratings of 4 and 5 were collapsed into 3’s (high). Using the collapsed 3-point scales to
estimate rWG resulted in values ranging from .53 to 1.00 with a mean of .79. These rWG values
along with the means and standard deviations of judges’ ratings are presented in Table 2 (conflict
response options) and Table 3 (helping response options). Using the recommendations provided
by LeBreton and Senter (2008) to interpret values of .51 to .70 as moderate agreement, .71 to .90
as strong agreement, and .91 to 1.00 as very strong agreement, 27.3%, 45.5%, and 27.3% the
rWG values fall within these categories, respectively (note that these percentages are computed
based on the 44 rWG values in Tables 2 and 3 combined). Thus, there was adequate agreement
among judges to use the means of ratings they provided.
Test takers’ likelihood ratings of SJT response options were consequently weighted by
the appropriate mean ratings provided by the sample of judges to come up with an assertiveness,
cooperativeness, and intervention score for each test taker for each of the response options
associated with the two conflict SJT items, and a helping score for each of the response options
associated with the two helping SJT items. Response option scores for each item were added up
to come up with an item-level score for each person (2 cooperativeness scores, 2 assertiveness
scores, and 2 intervention scores corresponding to the two conflict SJT items; 2 helping scores
corresponding to the two helping SJT items). Item-level scores were also summed up to come up
with test-level scores for test takers (i.e., overall scores for cooperativeness, assertiveness,
intervention, and helping).

64

Table 2. Means, standard deviations, and rWG values for judges' ratings of conflict response options
Dimension
Cooperativeness
Assertiveness
Intervention
Response Option
M SD rWG
M SD rWG M SD rWG
CS1 Option 1: Tell the coworker that you will continue to
2.55 .78
.69
1.88 .91 .58 1.08 .27 .96
watch for unglued cartons and get back to him/her.
CS1 Option 2: Assume that the cartons are being glued
2.55 .64
.80
1.10 .31 .95 1.11 .31 .95
properly because your coworker hasn't noticed the problem.
CS1 Option 3: Insist that there must be a problem your
coworker hasn't noticed since unglued cartons are arriving at
1.45 .75
.72
2.95 .32 .95 1.27 .61 .82
the warehouse.
CS1 Option 4: Ask another team member to discuss the
2.13 .76
.71
2.35 .74 .73 2.80 .41 .92
problem with your coworker.
CS1 Option 5: Let your team leader know that your coworker
1.30 .65
.79
2.85 .49 .88 2.70 .61 .82
is allowing unglued cartons to get to the warehouse.
CS2 Option 1: Ask your coworker what you could do to get
2.92 .35
.94
2.08 .97 .53 1.11 .40 .92
along better.
CS2 Option 2: Ask the team leader for suggestions about how
2.60 .67
.77
2.15 .89 .60 2.77 .54 .86
to work with your coworker.
CS2 Option 3: Try to avoid your coworker whenever possible
1.55 .68
.77
1.67 .84 .65 1.00 .00 1.00
because you can't change him/her.
CS2 Option 4: Tell your coworker that he/she should act like a
1.73 .93
.56
2.75 .63 .80 1.22 .48 .89
team member and try to work together.
CS2 Option 5: Ask the team leader to meet with you and your
2.43 .75
.72
2.53 .78 .69 2.95 .22 .98
coworker to work out your differences.
CS2 Option 6: Ask other team members for suggestions on
2.40 .84
.65
2.33 .83 .66 2.81 .40 .92
how to work with your coworker.
Note. CS1 = Conflict situation 1. CS2 = Conflict situation 2.

65

Table 3. Means, standard deviations, and rWG values for judges' ratings of helping response options
Helping Dimension
Response Option
HS1 Option 1: Explain to your coworker that you are very busy, but if he/she wants to talk later,
you'll tell him/her how you handle Frank.
HS1 Option 2: Suggest that your coworker ignore Frank because working with Frank can be difficult.
HS1 Option 3: Tell your coworker that he/she needs to learn how to handle his/her own conflicts with
Frank.
HS1 Option 4: Tell your coworker how you deal with Frank.
HS1 Option 5: Suggest that your coworker meet with Frank to talk about their differences.
HS1 Option 6: Suggest that your coworker talk to the team leader about his/her problem with Frank.
HS2 Option 1: Agree to switch with your coworker so that he/she can take the training course.
HS2 Option 2: Agree to switch with your coworker, but only if he/she will switch with you in a few
weeks.
HS2 Option 3: Tell your coworker you think he/she is taking advantage of the system and you won't
switch.
HS2 Option 4: Tell your coworker you won't switch, and he/she should wait until training is offered
at another time that doesn't interfere with his/her schedule.
HS2 Option 5: Tell your coworker that you think he/she is taking advantage of the system, but you
will help him/her find someone else to switch.
Note. HS1 = Helping situation 1. HS2 = Helping situation.

66

M

SD

rWG

2.30

.85

.64

1.43

.75

.72

1.30

.61

.82

2.97
2.80
2.50
2.70

.16
.52
.72
.65

.99
.87
.74
.79

2.95

.22

.98

1.33

.66

.78

1.63

.81

.67

2.33

.86

.63

As an alternative to collapsing the 5-point response scale into a 3-point scale to improve
agreement between judges’ ratings of the response options, I also tried to standardize ratings
within judge for each SJT item with rated dimension combination (e.g., for helping ratings given
to response options of helping item 1) before computing mean ratings. This procedure resulted in
adequate rWG values as seen in Tables 4 and 5. However, I proceeded to use the procedure
described earlier as it was more convenient to interpret means on a 3-point scale (as opposed to
means of Z-scores) in the context of substantive analyses of test takers’ likelihood ratings of
these response options. Additionally, using Z-score weights did not improve the reliability issue
described later. It should be noted that using this approach would eliminate support for
Hypothesis 3 (see ―H3, scoring method 1‖).

67

Table 4. Means, standard deviations, and rWG values for judges' standardized ratings of conflict response options
Dimension
Cooperativeness
Assertiveness
Intervention
Response Option
M
SD rWG
M
SD rWG M SD rWG
CS1 Option 1: Tell the coworker that you will continue to watch for
.54
.66
.78
-.29
.62 .81 -.71 .25 .97
unglued cartons and get back to him/her.
CS1 Option 2: Assume that the cartons are being glued properly because
.67
.87
.62 -1.28 .41 .92 -.69 .23 .97
your coworker hasn't noticed the problem.
CS1 Option 3: Insist that there must be a problem your coworker hasn't
-.56 .70
.76
.89
.34 .94 -.54 .40 .92
noticed since unglued cartons are arriving at the warehouse.
CS1 Option 4: Ask another team member to discuss the problem with
.11
.61
.81
.07
.50 .88 1.02 .39 .92
your coworker.
CS1 Option 5: Let your team leader know that your coworker is allowing
-.76 .58
.83
.63
.44 .90 .83 .72 .74
unglued cartons to get to the warehouse.
CS2 Option 1: Ask your coworker what you could do to get along better.
.99
.51
.87
-.22
.79 .69 -.85 .30 .96
CS2 Option 2: Ask the team leader for suggestions about how to work
.31
.47
.89
-.08
.73 .73 .66 .44 .90
with your coworker.
CS2 Option 3: Try to avoid your coworker whenever possible because
-.93 .79
.69
-.83 1.08 .42 -.96 .22 .98
you can't change him/her.
CS2 Option 4: Tell your coworker that he/she should act like a team
-.54 .91
.59
.68
.80 .68 -.74 .41 .92
member and try to work together.
CS2 Option 5: Ask the team leader to meet with you and your coworker
.08
.67
.78
.29
.72 .74 1.01 .30 .96
to work out your differences.
CS2 Option 6: Ask other team members for suggestions on how to work
.11
.66
.78
.13
.57 .84 .77 .36 .94
with your coworker.
Note. CS1 = Conflict situation 1. CS2 = Conflict situation 2.

68

Table 5. Means, standard deviations, and rWG values for judges' standardized ratings of helping response
options
Helping Dimension
Response Option
M
SD rWG
HS1 Option 1: Explain to your coworker that you are very busy, but if he/she wants to talk
.11
.75
.72
later, you'll tell him/her how you handle Frank.
HS1 Option 2: Suggest that your coworker ignore Frank because working with Frank can
-.82
.71
.75
be difficult.
HS1 Option 3: Tell your coworker that he/she needs to learn how to handle his/her own
-1.01 .63
.80
conflicts with Frank.
HS1 Option 4: Tell your coworker how you deal with Frank.
.76
.38
.93
HS1 Option 5: Suggest that your coworker meet with Frank to talk about their differences.
.71
.55
.85
HS1 Option 6: Suggest that your coworker talk to the team leader about his/her problem
.29
.50
.88
with Frank.
HS2 Option 1: Agree to switch with your coworker so that he/she can take the training
.54
.83
.66
course.
HS2 Option 2: Agree to switch with your coworker, but only if he/she will switch with you
.76
.36
.94
in a few weeks.
HS2 Option 3: Tell your coworker you think he/she is taking advantage of the system and
-.89
.64
.80
you won't switch.
HS2 Option 4: Tell your coworker you won't switch, and he/she should wait until training is
-.50
.68
.77
offered at another time that doesn't interfere with his/her schedule.
HS2 Option 5: Tell your coworker that you think he/she is taking advantage of the system,
.09
.67
.78
but you will help him/her find someone else to switch.
Note. HS1 = Helping situation 1. HS2 = Helping situation 2.

69

As an alternative to using means for weighting likelihood ratings assigned to response
options before computing composite scores, an attempt was made to instead apply unit weights
of 1 and 2 for low and high levels of a dimension of interest. Determinations of low or high were
made based on judges’ mean rating for a given response option falling either below or above the
midpoint on the collapsed 3-point rating scale. The weights assigned using this procedure are
presented in Tables 6 and 7 below. Unit weights like Z-score weights did not improve the
reliability issue described later and were ultimately not used. It should be noted, however, that
using this approach would produce the same results for Hypotheses 1 through 5.

70

Table 6. Unit weights associated with judges' standardized ratings of conflict response options
Response Option

Weight for
Cooperativeness

CS1 Option 1: Tell the coworker that you will continue to
watch for unglued cartons and get back to him/her.
CS1 Option 2: Assume that the cartons are being glued
properly because your coworker hasn't noticed the problem.
CS1 Option 3: Insist that there must be a problem your
coworker hasn't noticed since unglued cartons are arriving at
the warehouse.
CS1 Option 4: Ask another team member to discuss the
problem with your coworker.
CS1 Option 5: Let your team leader know that your
coworker is allowing unglued cartons to get to the
warehouse.
CS2 Option 1: Ask your coworker what you could do to get
along better.
CS2 Option 2: Ask the team leader for suggestions about
how to work with your coworker.
CS2 Option 3: Try to avoid your coworker whenever
possible because you can't change him/her.
CS2 Option 4: Tell your coworker that he/she should act like
a team member and try to work together.
CS2 Option 5: Ask the team leader to meet with you and
your coworker to work out your differences.
CS2 Option 6: Ask other team members for suggestions on
how to work with your coworker.

Weight for
Weight for
Assertiveness Intervention

2.00

1.00

1.00

1.00

2.00

1.00

2.00

2.00

2.00

1.00

2.00

2.00

2.00

2.00

1.00

2.00

2.00

2.00

1.00

1.00

1.00

1.00

2.00

1.00

2.00

2.00

2.00

2.00

71

1.00

2.00

Note. CS1 = Conflict situation 1. CS2 = Conflict situation 2.

1.00

2.00

2.00

Table 7. Unit weights associated with judges' standardized ratings of helping response options
Weight for
Helping

Response Option
HS1 Option 1: Explain to your coworker that you are very busy, but if he/she wants to talk
later, you'll tell him/her how you handle Frank.
HS1 Option 2: Suggest that your coworker ignore Frank because working with Frank can
be difficult.
HS1 Option 3: Tell your coworker that he/she needs to learn how to handle his/her own
conflicts with Frank.
HS1 Option 4: Tell your coworker how you deal with Frank.
HS1 Option 5: Suggest that your coworker meet with Frank to talk about their differences.
HS1 Option 6: Suggest that your coworker talk to the team leader about his/her problem
with Frank.
HS2 Option 1: Agree to switch with your coworker so that he/she can take the training
course.
HS2 Option 2: Agree to switch with your coworker, but only if he/she will switch with you
in a few weeks.
HS2 Option 3: Tell your coworker you think he/she is taking advantage of the system and
you won't switch.
HS2 Option 4: Tell your coworker you won't switch, and he/she should wait until training is
offered at another time that doesn't interfere with his/her schedule.
HS2 Option 5: Tell your coworker that you think he/she is taking advantage of the system,
but you will help him/her find someone else to switch.
Note. HS1 = Helping situation 1. HS2 = Helping situation 2.

72

2.00
1.00
1.00
2.00
2.00
2.00
2.00
2.00
1.00
1.00
2.00

Reliability of Measurement of Hypothesized SJT Item Dimensions
Reliability analysis performed on the weighted response options to verify that the
expected dimensions of helping, assertiveness, cooperativeness, and third party intervention were
being measured reliably by the appropriate weighted response options, showed internal
consistency reliability to be low in all cases. For the overall dimensions, alphas ranged from .404
for the cooperativeness dimension to .518 for the helping dimension. When reliability was
computed for the response options for each SJT item separately, reliability values ranged from
.257 for intervention based on one item to .604 for helping based on one item. These reliabilities
are shown in Table 8. Given that none of the alphas reached even the .70 minimum convention
of acceptability, there was inadequate evidence for the reliable measurement of the dimensions I
judged to be present in the response options. Given that the a priori dimensional structure was
not producing reliable measures, the likelihood ratings for the entire set of response options (i.e.,
22 options) was subjected to an exploratory factor analysis (EFA) to determine if a more
interpretable structure existed. Maximum likelihood estimation with varimax rotation was used
for the EFA. Eight factors were extracted and explained 38.38% of the variance. This model
adequately fit the data,

2

(84) = 90.04, ns. Loadings of items on the factors are presented in

Table 9. It should be noted that a forced solution of 4 factors (because there were four SJT items)
and 2 factors (because there were items of 2 types) were also tried but the goodness of fit test
2

[ (132) = 223.34, p < .001, and

2

(169) = 398.15, p < .001, respectively] indicated that these

models did not fit the data.
Three of the response options associated with helping situation 2 showed high loadings
onto the first factor, which explained 9% of the variance in the data. A close look at these three
response options confirmed that the factor could be considered ―helping‖. Likelihood ratings for
73

option 1, option 3 (reverse coded such that a higher likelihood rating corresponded to a higher
level of helping), and option 4 (also reverse coded) were therefore weighted by their associated
standing on the helping dimension (as rated by 40 judges) and added up to form a helping score.
Internal consistency reliability for this helping composite was .745. As can be seen in Table 9 an
additional response option (5

th

one) also loaded onto the first factor, however, it was not

included in the helping composite as it would reduce reliability down to .67. The remaining
items did not hang together strongly enough to warrant creating additional composite scores for
subsequent analyses (alphas of no higher than .64).

Table 8. Reliabilities of SJT item dimensions
Dimension
Alpha
Cooperativeness, overall
.404
Cooperativeness for conflict item 1
.047
Cooperativeness for conflict item 2
.407
Assertiveness, overall
.464
Assertiveness for conflict item 1
.261
Assertiveness for conflict item 2
.382
Intervention, overall
.467
Intervention for conflict item 1
.257
Intervention for conflict item 2
.409
Helping, overall
.518
Helping for helping item 1
.394
Helping for helping item 2
.604

74

Table 9. Loadings for likelihood ratings respondents gave to SJT response options
Factor
Rated Response Option
1
HS2 O1: Agree to switch with your coworker so that he/she can take the training
-.881
course.
HS2 O3: Tell your coworker you think he/she is taking advantage of the system
.708
and you won't switch.
HS2 O4: Tell your coworker you won't switch, and he/she should wait until
.683
training is offered at another time that doesn't interfere with his/her schedule.
HS2 O5: Tell your coworker that you think he/she is taking advantage of the
system, but you will help him/her find someone else to switch.
HS1 O6: Suggest that your coworker talk to the team leader about his/her
problem with Frank.
CS2 O5: Ask the team leader to meet with you and your coworker to work out
your differences.
CS2 O2: Ask the team leader for suggestions about how to work with your
coworker.
HS1 O5: Suggest that your coworker meet with Frank to talk about their
differences.
HS1 O1: Explain to your coworker that you are very busy, but if he/she wants to
talk later, you'll tell him/her how you handle Frank.
CS1 O1: Tell your coworker that you will continue to watch for unglued cartons
and get back to him/her.
HS2 O2: Agree to switch with your coworker, but only if he/she will switch
with you in a few weeks.

75

2

3

4

5

6

-.207

7

8
.290

.321

.242
.679
.534

.214

.509

.475

.377

-.230

-.227
.978

Table 9 (cont’d)
1
CS1 O5: Let your team leader know that your coworker is
allowing unglued cartons to get to the warehouse.
CS1 O4: Ask another team member to discuss the problem with
your coworker.
HS1 O3: Tell your coworker that he/she needs to learn how to
handle his/her own conflicts with Frank.
CS1 O3: Insist that there must be a problem your coworker
hasn't noticed since unglued cartons are arriving at the
warehouse.

3

4

5

6

7

8

.595
.455
-.210

.325
.294

HS1 O2: Suggest that your coworker ignore Frank because
working with Frank can be difficult.
CS2 O3: Try to avoid your coworker whenever possible
because you can't change him/her.
CS1 O2: Assume that the cartons are being glued properly
because your coworker hasn't noticed the problem.
CS2 O1: Ask your coworker what you could do to get along
better.
HS1 O4: Tell your coworker how you deal with Frank.
CS2 O6: Ask other team members for suggestions on how to
work with your coworker.
CS2 O4: Tell your coworker that he/she should act like a team
member and try to work together.
Variance explained

2

.271
.542
.530

-.213

.232
.623
.566
.408
.430
9.00% 6.14% 4.81% 4.57% 4.14% 3.73% 3.28% 2.72%

Note. N = 374. Factor loadings < .2 are suppressed. CS1 = Conflict situation 1. CS2 = Conflict situation 2. HS1 = Helping situation 1.
HS2 = Helping situation 2.
76

Recall that there are a number of ways the SJT in this study can be scored (or analyzed).
The first method ( the one most suited for testing the hypotheses in this study), is by weighting
likelihood ratings test takers gave to each response option according to the level of the construct
of interest they represent and summing these weighted ratings into a dimension score (e.g.,
helping score). The dimension score can then be used as the dependent variable. The second
method is to use the likelihood ratings test takers assigned to response options as is (i.e., not
weight them), as separate dependent variables. The third method is to score the most and least
likely responses test takers provided for each SJT item according to response effectiveness as
determined by the test developer. The dependent variable is then the item score (or the test score
if the four item scores are summed up).
Given the above discussion of reliability issues, scoring method 1 can only be applied for
testing some of the helping hypotheses (i. e., the ones that include or pertain uniquely to helping
situation 2 as none of the first helping situation’s response options went into forming the
―helping score‖). The hypotheses were therefore tested using scoring method 1 whenever
possible and explored (more so than tested) using the other two scoring methods when scoring
method 1 could not be applied at all or could only be partially applied. It should be pointed out,
however, that scoring method 3, as it is more appropriate for testing for effects on performance
rather than on constructs like helping or assertiveness, is not at all suitable for testing my
hypotheses pertaining to constructs. Method 3 is nonetheless applied and discussed in the
interests of understanding demographic cue effects on SJT performance.
Correlations
The intercorrelations, means, standard deviations, and reliabilities for study variables are
provided in Table 10 (note that explanation of variable names and other notes about the table are

77

presented on page 89). Inconsistent with the expectation that attractiveness of the non-focal
actors will influence responses to the SJT items, ratings of attractiveness were not related to
ratings of response options in expected ways. Thus, there was no strong support for controlling
for non-focal actor attractiveness in the analyses. Correlations of variables representing knowing
actors in the videos with responses on the SJT showed knowing actors to have very little
influence (knowing the non-focal actor in conflict situation 1 was related to likelihood ratings for
the first response option, r = -.12; knowing the non-focal actor in helping situation 1 was related
to likelihood ratings for the third response option, r = .13). Thus, it did not seem necessary to
control for knowing actors in the analyses.

78

SD
.81
1.05
1.05
1.13
1.19
1.10
1.08
1.12
1.09
1.29
1.10
1.34
1.02
.97
.71
.92
1.10

3. CS1.3

4. CS1.4

5. CS1.5

6. CS2.1

7. CS2.2

8. CS2.3

9. CS2.4

10. CS2.5

11. CS2.6

12. HS1.1

13. HS1.2

14. HS1.3

15. HS1.4

-.03
.01
-.02
-.03
.09
.11
-.10
-.02
.02
.03
.00
-.07
-.15
.07
.22
.18

-.07
-.02
-.17
-.04
-.09
.08
-.06
-.03
-.01
.12
.11
.06
.01
-.07
-.06

.14
.18
-.01
.06
.03
.20
.09
.10
.05
.00
.07
.15
.05
.07

.28
.13
.12
.04
.05
.10
.15
.11
.09
.16
-.07
.01
.07

.08
.10
-.05
.15
.19
.00
.06
.02
.17
-.07
.05
.09

.34
-.23
.08
.10
.12
.04
-.06
-.04
.02
.21
.00

-.07
.11
.36
.20
-.03
-.09
-.07
.09
.24
.32

-.01
-.13
.08
.03
.34
.05
.14
-.13
-.04

.21
.00
.03
-.10
.00
.10
.17
.09

.06
-.06
-.20
-.10
.06
.28
.36

.13
.04
.02
.21
-.01
.03

.16
.20
-.04
-.11
-.13

.17
.04
-.26
-.10

-.06
-.10
-.12

-.03
-.04

.28

19.23

5.62

.17

.04

-.16

-.10

-.14

.10

.03

-.05

-.10

-.04

-.01

-.12

-.05

-.18

.14

.11

3.56
4.16
2.57
2.88
2.86

1.16
.83
1.10
1.23
1.10

.16
.19
-.16
-.12
.00

.09
.05
-.04
.07
.06

-.16
.06
.12
.10
-.02

-.11
.01
.10
.04
.02

-.13
-.01
.15
.07
-.01

.11
.00
-.03
-.08
-.03

.08
-.01
.04
.01
.11

-.01
.03
.10
.06
.03

-.04
.06
.17
.10
.16

-.01
.03
.10
.03
.14

-.04
.12
-.01
-.01
.01

-.11
.03
.12
.08
.05

-.06
.06
.09
.00
-.04

-.16
-.03
.19
.13
-.06

.13
.13
-.11
-.10
.05

.14
.02
-.01
-.07
.01

79

16. HS1.5

M
4.29
2.51
3.48
2.77
3.03
3.92
3.72
2.48
3.52
3.27
3.69
2.58
1.94
1.89
4.21
4.05
3.66

2. CS1.2

1. CS1.1
2. CS1.2
3. CS1.3
4. CS1.4
5. CS1.5
6. CS2.1
7. CS2.2
8. CS2.3
9. CS2.4
10. CS2.5
11. CS2.6
12. HS1.1
13. HS1.2
14. HS1.3
15. HS1.4
16. HS1.5
17. HS1.6
18.
a
Helping
19. HS2.1
20. HS2.2
21. HS2.3
22. HS2.4
23. HS2.5

1. CS1.1

Table 10. Intercorrelations, means, standard deviations, and reliabilities

20. HS2.2

21. HS2.3

22. HS2.4

18. Helping
19. HS2.1
20. HS2.2
21. HS2.3
22. HS2.4
23. HS2.5

19. HS2.1

a

18. Helping

17. HS1.6

-.09
-.06
.02
.09
.09
.03

.75
.92
.16
-.76
-.82
-.23

.20
-.57
-.59
-.17

-.09
-.08
.01

.53
.28

.18

23. HS2.5

17. HS1.6

Table 10 (cont’d)

-

80

14. HS1.3

15. HS1.4

16. HS1.5

13. HS1.2

.03 .06 .01 -.06 -.05
.38 .17 .01 -.26 -.04
.03 -.07 .00 -.32 .01
-.01 .01 -.06 -.11 -.07
.20 .07 -.01 -.35 -.06
.07 .03 -.01 -.03 .00
.06 .00 .02 .02 .03
.03 .03 -.03 .01 .09
.08 .09 -.02 -.07 .03
.08 .06 .00 .00 .04
-.04 .01 .01 -.04 .02
-.03 .04 .06 -.02 -.06
-.11 -.02 .01 -.04 .00
-.04 .00 .01 .04 -.05

.11
.01
-.11
.04
.02
-.01
.06
.00
.04
.07
.04
.02
.01
.02

.07
.22
.33
.12
.35
.14
.04
.06
.12
.06
.06
-.04
-.03
.04

.06

.03

-.01

.03

.01

.04

.05

.02

.08

.06

-.02

.07

.06

.10

.01

.01

.07

-.01

.01

-.03

.01

-.04

.07

.00

.00

.03

-.02

.05

-.03

.01

-.01

.03

-.06

.00

-.04

.05

.01

-.03

.02

-.03

.00

-.03

.04

.01

-.03

.10

.02

-.01

.00

.01

.00

.02

.07

.03

.00

.10

-.02

.13

.02

.00

.06

-.01 -.03 -.08 -.04

.02

.03

.01

.05

.01

-.02 -.04

.09

-.05

-.03

81

12. HS1.1

.00
.14
.09
-.07
.09
.06
.01
.02
.01
.02
-.12
-.04
-.03
-.13

11. CS2.6

9. CS2.4

-.27 .00 .01 -.04
.11 .18 .29 -.37
-.03 .11 .00 -.17
-.10 .01 .07 -.06
-.14 .15 .16 -.30
.10 .14 .20 -.05
.08 .09 .13 .01
.04 .11 .09 -.10
.04 .13 .07 -.19
.06 .05 .05 -.16
-.09 -.03 -.02 .02
-.04 -.07 .02 .04
-.06 .02 -.01 .01
-.03 .01 -.01 -.01

10. CS2.5

8. CS2.3

.03
.06
-.02
.02
.03
.11
.10
.08
-.07
-.02
-.03
-.03
-.05
.00

7. CS2.2

.15
.09
.02
-.02
.12
.04
.01
.01
-.05
-.03
-.06
-.17
-.09
-.04

6. CS2.1

4. CS1.4

-.11
-.08
.08
.03
-.04
-.05
.00
-.04
-.09
-.03
.03
.01
.15
-.02

5. CS1.5

3. CS1.3

M
SD
24. CS1 SJT score 1.17 1.00 .17
25. CS2 SJT score
.55 .85 .06
26. HS1 SJT score
.59 .97 -.03
27. HS2 SJT score
.62 .60 .19
28. Total SJT score 2.93 1.87 .16
29. Opp Perform
2.94 .93 .06
30. Job Rel. Pre
2.00 .91 -.03
31. Job Rel. Con
2.15 .96 .01
32. Attraction
3.62 .83 .22
33. Compatibility
3.15 .90 .15
34. Att_CS1
2.93 .91 .05
35. Att_CS2
3.12 .98 .04
36. Att_HS1
2.97 .99 -.03
37. Att_HS2
2.92 .98 .06
38. Acq_CS1.focal
.02 .14 .00
(0=No; 1=Yes)
39. Acq_CS1.non.03 .16 -.12
focal (0=N; 1=Y)
40. Acq_HS1.focal
.02 .14 -.02
(0=No; 1=Yes)
41. Acq_HS1.non.01 .09 -.11
focal (0=N; 1=Y)
42. Acq_CS2.focal
.02 .15 -.08
(0=No; 1=Yes)

2. CS1.2

1. CS1.1

Table 10 (cont’d)

17. HS1.6

19. HS2.1

20. HS2.2

21. HS2.3

22. HS2.4

23. HS2.5

24. CS1 SJT score
25. CS2 SJT score
26. HS1 SJT score
27. HS2 SJT score
28. Total SJT score
29. Opp Perform
30. Job Related Pre
31. Job Related Con
32. Attraction
33. Compatibility
34. Att_CS1
35. Att_CS2
36. Att_HS1
37. Att_HS2
38. Acq_CS1.focal (0=No; 1=Yes)
39. Acq_CS1.nonfocal (0=No; 1=Yes)
40. Acq_HS1.focal (0=No; 1=Yes)
41. Acq_HS1.nonfocal (0=No; 1=Yes)
42. Acq_CS2.focal (0=No; 1=Yes)

18.
Helping

Table 10 (cont’d)

.02
.23
-.05
.09
.12
.02
.05
.03
.07
.06
.02
.01
-.03
.00
-.03
-.04
-.01
-.05
-.05

.11
-.06
.13
.17
.15
-.05
.00
.02
.11
.07
.12
.06
.14
.01
.04
.00
.05
.07
-.02

.09
-.07
.16
.18
.16
.02
.03
.06
.13
.07
.11
.07
.16
.01
.05
-.01
.05
.06
.00

.13
-.05
.06
.11
.11
.06
.04
.01
.13
.08
-.06
.02
.01
.04
.05
-.05
.02
-.02
-.07

-.12
.04
-.06
-.25
-.15
.11
.07
.08
-.04
.02
-.15
-.06
-.12
-.04
-.02
.02
-.05
-.02
.00

-.09
.03
-.06
-.01
-.07
.11
-.01
-.01
-.08
-.09
-.04
-.01
-.05
.03
-.02
-.02
-.02
-.09
.04

-.05
.04
.00
-.03
-.02
.12
.04
.04
-.05
.03
-.04
.00
.00
.02
.04
.07
.09
.04
.07

82

8. CS2.3

9. CS2.4

10. CS2.5

14. HS1.3

15. HS1.4

16. HS1.5

.08

.01

.00

-.04 -.01 -.03

.08

-.03

.01

-.01 -.05 -.02 -.01

.00

-.03

.01

.07

-.07

.10

.00

-.02

.00

-.01

.03

.07

.04

.02

.05

-.03

.08

.03

.00

.03

.18

-.01

.03

-.02

.00

-.04 -.08 -.06

.05

-.02 -.01 -.09 -.06

.04

-.04 -.06 -.04

.15

.63

-.08

.08

.03

-.03 -.03 -.05

.04

.02

.04

-.02 -.01 -.01

.07

-.01 -.03

.49

.50

-.03

.01

-.03

.03

-.05 -.08 -.13 -.01

.00

-.05

.05

.06

.13

-.06 -.09

.78

.42

.05

-.06 -.15 -.09 -.03

.06

-.07 -.04

.02

-.10 -.07 -.03

19.97

2.68

.01

-.02 -.03 -.07

-.06 -.08 -.04 -.10

2.39

1.08 -.03 -.01 -.07 -.09 -.02 -.06 -.02

.00

-.06 -.01

.04

.03

.00

-.02 -.05 -.04

5.95

1.89

.09

.00

.08

-.06

.02

-.05

.00

13. HS1.2

7. CS2.2

.02

.04

12. HS1.1

4. CS1.4

.12

11. CS2.6

3. CS1.3

.01

6. CS2.1

2. CS1.2

43. Acq_CS2.
non-focal (0=N;
1=Y)
44. Acq_HS2.
focal (0=N; 1=Y)
45.
Acq_HS2.nonfocal (0=N; 1=Y)
46. # Actor(s)
knew
47. Gender
(0=Female;
1=Male)

SD

5. CS1.5

M

1. CS1.1

Table 10 (cont’d)

.06

b

48. Race
(0=Minority;
1=White)
49. Age
50. Year in
college
c
51. GPA

.13

-.01 -.05

.01

.01

-.02

.08

83

.03

.09

.04

.04

.01

.02

-.02

-.05 -.06 -.01 -.05 -.07 -.04

.04

.09

21. HS2.3

-.02

-.04

.03

.06

.05

.06

.03

.00

-.05

.04

-.05

-.02

.00

.05

.08

.06

.02

.03

-.02

-.01

.01

.10

-.06

-.11

-.10

.01

-.01

-.18

.05

.05

.04

.07

-.02

-.05

.00

.05
.01
.05

.01
-.02
.16

-.04
-.06
.15

-.01
-.04
.12

-.08
-.05
-.11

-.03
-.01
-.13

-.13
-.13
-.01

23. HS2.5

20. HS2.2

-.02

43. Acq_CS2.nonfocal
.06
(0=No; 1=Yes)
44. Acq_HS2.focal
-.04
(0=No; 1=Yes)
45. Acq_HS2.nonfocal
.03
(0=No; 1=Yes)
46. # Actor(s) knew
-.02
47. Gender (0=Female;
-.09
1=Male)

22. HS2.4

19. HS2.1

-.01

17. HS1.6

18. Helping

Table 10 (cont’d)

b

48. Race
(0=Minority; 1=White)
49. Age
50. Year in college
c

51. GPA

84

32. Attraction

33. Compatibility

34. Att_CS1

35. Att_CS2

36. Att_HS1

37. Att_HS2

.03
.03
.08
.59
.07
-.02
-.03
.11
.07
.06
-.03
.01
-.02

.06
.07
.53
.17
.02
.10
.13
.12
.01
-.01
-.10
-.09

.08
.59
.07
.00
.07
.10
.11
-.01
-.01
.07
.03

.44
.01
-.02
-.01
.17
.10
.08
-.01
-.03
.02

.15
-.01
.06
.23
.18
.05
-.03
-.02
-.03

.84
.48
.39
.14
.14
.00
.05
.06
.06

.81
.54
.15
.15
-.01
.06
-.01
-.02

.83
.15
.11
-.04
.02
.03
.02

.88
.75
.06
.16
.04
.05

.85
-.06
.08
-.04
.04

.36
.48
.23

.35
.23

.32

-

-.04

.03

-.04

-.04

-.05

.12

.03

.02

.06

.00

.01

.06

.06

.03

-

.00

.11

.02

.02

.07

.01

-.07

.04

.05

.07

.00

.10

.04

-.02

.22

-

.00

-.07

-.06

.05

-.05

-.08

-.02

-.04

-.06

.00

.06

.08

.00

.01

.13

.22

-

.01

-.02

-.02

-.04

-.03

.04

.02

.03

.04

.03

-.03

.08

.09

-.05

.43

.36

.43

24. CS1 SJT
score
25. CS2 SJT
score
26. HS1 SJT
score
27. HS2 SJT
score
28. Total SJT
score
29. Opp
Perform
30. Job Rel.
Pre
31. Job Rel.
Con
24. CS1 SJT score
25. CS2 SJT score
26. HS1 SJT score
27. HS2 SJT score
28. Total SJT score
29. Opp Perform
30. Job Rel. Pre
31. Job Rel. Con
32. Attraction
33. Compatibility
34. Att_CS1
35. Att_CS2
36. Att_HS1
37. Att_HS2
38. Acq_CS1.focal
(0=N; 1=Y)
39. Acq_CS1.nonfocal (0=N; 1=Y)
40. Acq_HS1.focal
(0=N; 1=Y)
41. Acq_HS1.nonfocal (0=N; 1=Y)

38. Acq_CS1.
focal
39. Acq_CS1.
nonfocal
40. Acq_HS1.
focal
41. Acq_HS1.
nonfocal

Table 10 (cont’d)

85

-

30. Job Rel.
Pre

31. Job Rel.
Con

32. Attraction

33. Compatibility

34. Att_CS1

35. Att_CS2

36. Att_HS1

37. Att_HS2

c

51. GPA

26. HS1 SJT
score
27. HS2 SJT
score
28. Total SJT
score
29. Opp
Perform

42. Acq_CS2.
focal (0=N; 1=Y)
43. Acq_CS2.nonfocal (0=N; 1=Y)
44. Acq_HS2.
focal (0=N; 1=Y)
45. Acq_HS2.nonfocal (0=N; 1=Y)
46. # Actor(s)
knew
47. Gender
(0=Female;
1=Male)
b
48. Race
(0=Minority;
1=White)
49. Age
50. Year in college

25. CS2 SJT
score

40. Acq_HS1.
focal

24. CS1 SJT
score

38. Acq_CS1.
focal
39. Acq_CS1.
nonfocal

Table 10 (cont’d)

-.10

-.04

-.02

.04

-.07

.00

.01

.01

-.03

-.06

-.04

.00

.00

-.04

.37

.08

.37

.00

-.02

-.07

.07

-.02

-.05

-.01

.02

.07

.08

-.04

.06

-.07

-.04

.16

.27

.16

-.01

.04

-.01

-.01

.00

.00

-.06

.01

.05

.02

-.03

.10

.11

-.07

.53

.44

.26

.04

-.07

-.09

.07

-.03

.15

.05

.05

.02

.04

-.03

-.02

.02

.08

.19

.24

.19

-.02

-.01

-.07

.04

-.04

.05

-.01

.03

.04

.03

-.02

.08

.05

-.01

.60

.59

.56

.03

.02

-.07

-.08

-.03

.00

-.02

.08

.09

.17

-.17

-.17

-.13

-.13

-.02

.03

-.02

.04

.01

.02

.14

.08

-.13

-.14

-.13

.00

.04

-.01

.06

.06

-.03

-.12

-.16

-.02

.03
-.06

.02
-.01

-.13
-.16

-.04
-.05

-.06
-.14

-.06
-.16

-.01
-.06

.00
-.08

.02
.04

-.01
.03

-.05
-.01

.00
.02

-.02
-.05

.07
.10

.01
.00

.01
-.03

.08
.10

.03

.01

.06

.09

.08

.03

.04

.05

-.05

-.11

.03

.03

.02

-.07

.05

-.03

.06

86

42. Acq_CS2.focal (0=No; 1=Yes)
43. Acq_CS2.nonfocal (0=No; 1=Yes)
44. Acq_HS2.focal (0=No; 1=Yes)
45. Acq_HS2.nonfocal (0=No; 1=Yes)
46. # Actor(s) knew
47. Gender (0=Female; 1=Male)

51. GPA

50. Year in college

49. Age

48. Race (0,1)

47. Gender (0,1)

46. # Actor(s) knew

45. Acq_HS2.nonfocal

44. Acq_HS2.focal

43. Acq_CS2.nonfocal

42. Acq_CS2.focal

41. Acq_HS1.nonfocal

Table 10 (cont’d)

.38
.25 .13
.82 .23 .31
.31 .16 .36
.19
.74 .57 .53
.68
.61
.09 .09 .02
.07
-.01
.05
b
-.10 .00 -.01 -.05 -.09 -.12 -.05
48. Race (0=Minority; 1=White)
49. Age
.03 .02 .03 -.01
.00
.04
.20
-.11
50. Year in college
-.03 -.02 .02 -.06
.01
.00
-.09
.12
.55
c
.00 .05 -.11 .02
-.10 -.02 -.12
.19
-.14 -.17
51. GPA
Note. Correlations in bold are significant at at least the .05 level (2-tailed). Ns range from 348 (for race) to 374. CS1= conflict
situation 1. CS2= conflict situation 2. HS1= helping situation 1. HS2= helping situation 2. Att_CS1, Att_CS2, Att_HS1, and Att_HS2
refer to the rated attractiveness of the non-focal actor in conflict situation 1, conflict situation 2, helping situation 1, and helping
situation 2, respectively. Variables starting with "Acq" refer to whether or not the respondent was acquainted with the focal or nona
b
focal actor in a particular SJT item. Helping [score] is composed of HS2.1, HS2.3, and HS2.4. Race variable excludes those who
c

were multiracial or "other." GPA was a categorical variable.

87

Tests of Hypotheses Pertaining to Helping
Before discussing the hypotheses pertaining to helping situations, I include Tables 11 and
12 below to provide an overview of mean likelihood ratings as a function of sex (Table 11) and
race (Table 12) of the respondent and non-focal actor.

88

Table 11. Likelihood ratings for helping response options as a function of respondent's and non-focal actor's sex
Female Respondent
Male Respondent
Female nonMale nonFemale nonMale nonfocal actor
focal actor
focal actor
focal actor
(N = 96)
(N = 93)
(N = 95)
(N = 90)
Response Option
M
SD
M
SD
M
SD
M
SD
HS1.1: Explain to your coworker that you are very busy, but if
2.69
1.48
2.31
1.24
2.87
1.28
2.42 1.27
he/she wants to talk later, you'll tell him/her how you handle Frank.
HS1.2: Suggest that your coworker ignore Frank because working
2.01
1.05
1.74
1.01
2.05
1.00
1.93
.99
with Frank can be difficult.
HS1.3: Tell your coworker that he/she needs to learn how to handle
1.74
1.00
1.80
.85
2.04
1.08
1.98
.92
his/her own conflicts with Frank.
HS1.4: Tell your coworker how you deal with Frank.
4.24
.56
4.27
.74
4.26
.64
4.08
.86
HS1.5: Suggest that your coworker meet with Frank to talk about
4.17
.85
4.09
.88
4.04
.96
3.88
.98
their differences.
HS1.6: Suggest that your coworker talk to the team leader about
3.77
1.11
3.74
1.01
3.54
1.14
3.59 1.13
his/her problem with Frank.
HS2.1: Agree to switch with your coworker so that he/she can take
3.69
1.14
3.69
1.08
3.67
1.15
3.17 1.20
the training course.
HS2.2: Agree to switch with your coworker, but only if he/she will
switch with you in a few weeks.
HS2.3: Tell your coworker you think he/she is taking advantage of
the system and you won't switch.
HS2.4: Tell your coworker you won't switch, and he/she should
wait until training is offered at another time that doesn't interfere
with his/her schedule.
HS2.5: Tell your coworker that you think he/she is taking
advantage of the system, but you will help him/her find someone
else to switch.
Note. HS1 = Helping situation 1. HS2 = Helping situation 2.
89

4.15

.85

4.33

.86

4.18

.68

3.96

.90

2.52

1.10

2.59

1.13

2.43

1.05

2.74

1.11

2.81

1.23

2.96

1.36

2.65

1.16

3.10

1.13

3.06

1.00

3.05

1.11

2.46

1.09

2.88

1.10

Table 12. Likelihood ratings for helping response options as a function of respondent's and non-focal actor's race
Asian Respondent
Black Respondent
White Respondent
Asian nonfocal actor
(N = 17)

Black nonfocal actor
(N = 12)

White nonfocal actor
(N = 12)

Asian nonfocal actor
(N = 11)

Black nonfocal actor
(N = 14)

White nonfocal actor
(N = 10)

Asian nonfocal actor
(N = 94)

Black nonfocal actor
(N = 89)

White nonfocal actor
(N = 88)

M

SD

M

SD

M

SD

M

SD

M

SD

M

SD

M

SD

M

SD

M

SD

HS1.1

3.47

1.33

3.00

1.65

2.67

1.30

3.36

1.36

2.29

1.07

2.20

1.03

2.34

1.30

2.62

1.34

2.66

1.37

HS1.2

2.12

.93

2.75

1.29

1.92

1.00

2.27

1.10

1.57

.85

1.70

1.06

1.94

1.01

1.93

.99

1.86

1.01

HS1.3

2.35

1.37

1.58

.67

2.42

1.08

1.91

1.22

1.64

1.01

1.50

.97

1.66

.78

1.85

.94

2.11

1.01

HS1.4

4.06

.75

4.25

.87

4.67

.49

4.00

.45

4.07

.73

4.10

.57

4.20

.61

4.33

.65

4.15

.84

HS1.5

4.12

.93

3.75

1.06

3.92

1.00

4.09

.94

4.50

.65

4.00

.94

4.16

.79

3.98

.98

3.98

.97

HS1.6

3.65

.93

3.42

.90

2.92

1.51

3.82

1.08

3.71

1.07

3.80

1.14

3.83

1.09

3.70

1.10

3.52

1.12

HS2.1

3.29

1.16

3.33

1.30

4.08

1.08

2.45

1.04

3.93

.92

3.80

.92

3.51

1.10

3.48

1.24

3.81

1.07

HS2.2

4.24

.97

4.17

.72

3.83

1.19

3.91

.94

4.21

.70

3.80

1.14

4.17

.80

4.27

.75

4.16

.80

HS2.3

2.53

1.12

2.42

1.56

2.17

.94

3.36

1.03

2.36

1.22

2.90

.88

2.53

1.04

2.62

1.02

2.41

1.14

HS2.4

3.00

1.27

3.00

1.28

2.17

1.19

3.73

.79

3.21

1.48

2.90

1.29

2.83

1.26

2.82

1.15

2.83

1.23

HS2.5

2.94

1.03

2.67

1.30

2.33

.98

3.00

1.00

2.79

1.19

3.60

1.07

2.91

1.11

2.83

1.04

2.81

1.08

Note. HS1 = Helping situation 1. HS2 = Helping situation 2.

90

H1, scoring method 1. Hypothesis 1 stated that respondent race will moderate the effects
of the race mismatch between the respondent and the non-focal actor on helping, such that the
effects of race mismatch will be more negative for White respondents viewing a minority nonfocal actor than for minority respondents viewing a White non-focal actor. To test this
hypothesis, an analysis of variance (ANOVA) was conducted with dichotomized respondent race
(White, minority) and race match as the independent variables. The ANOVA showed no
significant effect of the race by race match interaction on helping, F(1,316) = .02, MSE = 31.44,
ns. Thus, Hypothesis 1 was not supported using scoring method 1.
H1, scoring method 2. Additionally, a MANOVA was calculated to examine the effects
of race mismatch and respondent race on likelihood ratings respondents provided for the eleven
helping response options. These were entered in as a set of dependent variables into the
MANOVA. There was no significant effect of the respondent race by race match interaction on
likelihood ratings on the set of response options, Wilks’ lambda = .96, F(11,296) = 1.13, ns.
Thus, Hypothesis 1 was not supported using scoring method 2.
H1, scoring method 3. Scoring method 3 was applied to examine the effects of race
mismatch and respondent race on effectiveness of responses to helping situations 1 and 2. The
interaction of race mismatch and dichotomized respondent race on helping situation 1 score was
not statistically significant, F(1,329) = .97, MSE = .92, ns. The main effect of race mismatch was
marginally significant, F(1,329) = 3.76, MSE = .92, p = .053. Respondents scored higher on
helping situation 1 when the non-focal actor was of the same race as them (M = .71, SD = .94) as
opposed to when there was a race mismatch (M = .53, SD = .96).
Next, looking at helping situation 2, the interaction of race mismatch and dichotomized
respondent race on helping situation 2 score was significant, F(1,316) = 9.52, MSE = .33, p <

91

.01, as were the main effects for race mismatch, F(1,316) = 9.13, MSE = .33, p < .01, and for
dichotomized respondent race, F(1,316) = 6.39, p < .05. White respondents scored the same on
helping item 2 regardless of whether there was a race match (M = .68, SD = .52) or a race
mismatch (M = .68, SD = .59) with the non-focal actor, F(1,250) = .01, MSE = .32, ns. Minority
respondents, however, scored higher on average on helping item 2 when the non-focal actor was
of the same race as them (M = .77, SD = .65) than when there was a race mismatch between the
respondent and non-focal actor (M = .23, SD = .58), F(1, 66) = 10.81, MSE = .37, p < .01. Thus,
with respect to response effectiveness, there were (marginally) negative effects of race mismatch
on both minority and White respondents for helping situation 1 and more negative effects of race
mismatch on minority than on White respondents for helping situation 2.
To find a substantive explanation for the significant effect of race mismatch on minority
respondents’ helping situation 2 scores, cross-tabs with a chi-square test of independence were
calculated on their ML and LL responses with the goal of examining how the frequencies with
which different options were selected as most or least likely were affected by race match with the
non-focal actor. Race match did not have a statistically significant effect on the frequency with
which various response options were selected as most likely,
likely,

2

(3, N = 333) = 5.51, ns, or least

2

(3, N = 333) = 4.97, ns. Therefore, actual frequencies of responses were not further

examined.
H2, scoring method 1. To test Hypothesis 2, that there will be a main effect for
respondent sex on helping, such that female respondents will offer more help to non-focal actors
than will male respondents, an ANOVA was performed. The effect of gender was not significant,
F(1,345) = 1.51, MSE = 31.73, ns. Thus, Hypothesis 2 was not supported using scoring method
1.
92

H2, scoring method 2. Additionally, a MANOVA was calculated to examine the effect
of respondent sex on likelihood ratings respondents provided for the entire set of helping
response options. The MANOVA showed a significant effect of respondent sex on the set of
dependent variables, Wilks’ lambda = .93, F(11,325) = 2.39, p < .01. Follow up analyses
showed that female respondents gave significantly lower likelihood ratings (M = 1.79, SD = .94)
than male respondents (M = 2.02, SD = .99) to helping situation 1’s response option 3 (Tell your
coworker that he/she needs to learn how to handle his/her own conflicts with Frank), t(335) = 2.18, p < .05. Response option 3 was rated by judges as low on helping (M = 1.30 on 3-point
scale, SD = .61), indicating that women’s lower average likelihood rating for this option can be
said to correspond to a higher willingness (relative to male respondents) to help the non-focal
actor. Female respondents also gave significantly higher likelihood ratings (M = 3.03, SD = 1.06)
than male respondents (M = 2.64, SD = 1.12) to helping situation 2’s response option 5 (Tell
your coworker that you think he/she is taking advantage of the system, but you will help him/her
find someone else to switch), t(335) = 3.30, p = .001. Response option 5 was rated by judges as
above average on helping (M = 2.33 on a 3-point scale, SD = .86), suggesting that women’s
higher average likelihood rating for this option (relative to men’s) should correspond to a higher
willingness to help the non-focal actor. Finally, women also gave somewhat higher likelihood
ratings than men to option 6 of helping situation 1 (Suggest that your coworker talk to the team
leader about his/her problem with Frank), t(335) = 1.81, p = .072, and option 2 of helping
situation 2 (Agree to switch with your coworker, but only if he/she will switch with you in a few
weeks), t(335) = 1.91, p = .057. Female respondents’ higher average ratings on these options (M
= 3.74 with SD = 1.09 and M = 4.26 with SD = .85, respectively) relative to male respondents’
(M = 3.52 with SD = 1.15 and M = 4.09 with SD = .80, respectively) should correspond to a

93

higher willingness to help the non-focal actor given that judges rated these two response options
highly on helping (M = 2.5 with SD = .72 and M = 2.95 with SD = .22, respectively). Thus,
Hypothesis 2 was partially supported using scoring method 2.
H2, scoring method 3. Next, scoring method 3 was applied to examine the effects of
respondent sex on effectiveness of responses to helping situations 1 and 2. There was no
statistically significant main effect for respondent sex on helping situation 1 score, F(1,360) =
1.42, MSE = .94, ns. The main effect of respondent sex on helping situation 2 score was
marginally significant, F(1,345) = 2.85, MSE = .36, p = .092. Women scored somewhat higher
(M = .68, SD = .59) on helping situation 2 than men (M = .57, SD = .61). Thus, with respect to
response effectiveness, there were no significant differences between male and female
respondents.
H3, scoring method 1. Hypothesis 3 stated that respondent sex will moderate the effects
of non-focal actor sex on helping, such that a female non-focal actor will receive more help than
a male non-focal actor when the respondent is male while female and male non-focal actors will
receive equal amounts of help when the respondent is female. An ANOVA showed a significant
effect of the non-focal actor sex by respondent sex interaction on helping, F(1,343) = 3.93, MSE
= 30.98, p < .05. The main effect for non-focal actor sex was also significant, F(1,343) = 6.29,
MSE = 30.98, p < .05. To break down the significant two way interaction, separate ANOVAs for
male and female respondents were calculated and revealed a statistically significant simple main
effect of non-focal actor sex for male respondents, F(1,173) = 10.62, MSE = 29.61, p = .001, but
not for female respondents, F(1,170) = .13, MSE = 32.39, ns. Male respondents were willing to
offer more help on average to female (M = 20.17, SD = 5.28) than to male coworkers (M =
17.49, SD = 5.61). Female respondents were willing to offer equal help to female (M = 19.81, SD

94

= 5.58) and male coworkers (M = 19.50, SD = 5.80). Thus, Hypothesis 3 was supported using
scoring method 1.
H3, scoring method 2. As an additional test of Hypothesis 3 at the level of the helping
response options, a MANOVA was calculated with non-focal actor sex and respondent sex as
independent variables. The interaction of the two independent variables was not statistically
significant, Wilks’ lambda = .95, F(11, 323) = 1.43, ns. This means that Hypothesis 3 did not
receive support at the level of the individual response options. Consistent with the earlier
analyses with regard to Hypothesis 2 though, there was a significant main effect for respondent
gender, Wilks’ lambda = .93, F(11,323) = 2.37, p < .01. The main effect for non-focal actor
gender was also significant, Wilks’ lambda = .93, F(11,323) = 2.28, p < .05. On average,
respondents gave helping situation one’s options 1 and 2 higher likelihood ratings when the nonfocal actor was female, t(359.06) = 2.59, p = .01, and t(360) = 1.91, p = .057, respectively.
Respondents also gave lower ratings to helping situation two’s response options 4 and 5 when
the non-focal actor was female, t(345) = -2.57, p < .05, and t(345) = -1.90, p = .058, respectively.
The means and standard deviations by non-focal actor gender for likelihood ratings on these
three response options are shown in Table 13. Response option 1 (Explain to your coworker that
you are very busy, but if he/she wants to talk later, you'll tell him/her how you handle Frank) is
on the higher end of helping (M = 2.30 on a 3-point scale, SD = .85), response option 2 (Suggest
that your coworker ignore Frank because working with Frank can be difficult) is on the lower
end (M = 1.43 on a 3-point scale, SD = .75), response option 4 (Tell your coworker you won't
switch, and he/she should wait until training is offered at another time that doesn't interfere with
his/her schedule) is on the lower end (M = 1.63 on a 3-point scale, SD = .81) and option 5 (Tell
your coworker that you think he/she is taking advantage of the system, but you will help him/her

95

find someone else to switch) is on the higher end (M = 2.33 on a 3-point scale, SD = .86). In the
context of the mean differences in likelihood ratings this suggests that respondents were more
willing to help female non-focal actors in the case of response options 1 and 4 but were
somewhat more willing to help male non-focal actors in the cases of response options 2 and 5.

Table 13. Likelihood ratings for three response options as a function of non-focal actor sex
Female non-focal actor
Male non-focal actor
M
SD
N
M
SD
N
Rated Response Option
HS1 Op1: Explain to your coworker that
you are very busy, but if he/she wants to
185
2.39
1.26
177
talk later, you'll tell him/her how you handle 2.75 1.39
Frank.
HS1 Op2: Suggest that your coworker
ignore Frank because working with Frank
2.02 1.03
185
1.82
.99
177
can be difficult.
HS2 Op4: Tell your coworker you won't
switch, and he/she should wait until training
2.69 1.18
180
3.03
1.25
167
is offered at another time that doesn't
interfere with his/her schedule.
HS2 Op5: Tell your coworker that you think
he/she is taking advantage of the system, but
2.73 1.09
180
2.95
1.10
167
you will help him/her find someone else to
switch.
Note. HS1 = Helping situation 1. HS2 = Helping situation 2.

H3, scoring method 3. Next, scoring method 3 was applied to examine the effects of
non-focal actor sex and respondent sex on effectiveness of responses to helping situations 1 and
2. The interaction of non-focal actor sex and respondent sex on helping situation 1 score and on
helping situation 2 score was not statistically significant in both cases, F(1,358) = .37, MSE =
.94, ns, and F(1,343) = .52, MSE = .36, ns, respectively. Consistent with analyses reported
earlier, there was a marginally significant main effect of respondent sex on helping situation 2

96

score, F(1,343) = 2.99, MSE = .36, p = .085, with women scoring slightly higher than men on the
item.
H4, scoring method 2. Hypothesis 4, which stated that in a helping situation where the
competence stereotype is activated, the lowest level of help will be offered by White male
respondents to male Black non-focal actors as compared to other race and sex-based
combinations of respondents with targets, could not be tested as planned using scoring method 1.
Method 2 was used to do the analysis at the level of the response options associated with helping
situation 1 (the relevant situation for this hypothesis). A MANOVA with respondent sex (male,
female), dichotomized respondent race (minority, White), non-focal actor sex (male, female),
and non-focal actor race (Black, White) as the independent variables and likelihood ratings for
the six response options associated with helping situation 1 as the dependent variables was
calculated. Table 14 displays the 16 groups that result from the four way interaction of the
independent variables along with the associated sample sizes. Notably, Asian non-focal actors
were not included in this analysis. Consistent with earlier analyses, there was a main effect for
respondent gender, Wilks’ lambda = .91, F(6,198) = 3.26, p < .01. The MANOVA did not show
a statistically significant four-way interaction, Wilks’ lambda = .99, F(6,198) = .34, ns. The
results at the response option level, therefore, do not lend support to Hypothesis 4. It should be
noted though that given sample size limitations the power for finding the effect was very low.

97

Table 14. Groups compared in testing Hypothesis 4
Respondent
Respondent
Non-focal Actor
Gender
Race
Gender
Minority
(Asian, Black)

Female coworker
Male coworker

Female
Female coworker
White
Male coworker

Minority
(Asian, Black)

Female coworker
Male coworker

Male
Female coworker
White
Male coworker

Non-focal Actor
Race
Black coworker
White coworker
Black coworker
White coworker
Black coworker
White coworker
Black coworker
White coworker
Black coworker
White coworker
Black coworker
White coworker
Black coworker
White coworker
Black coworker
White coworker

N
5
5
7
5
26
21
20
21
11
4
3
9
19
23
23
17

H4, scoring method 3. Next, scoring method 3 was applied to examine the effects of the
four way interaction of respondent sex, dichotomized respondent race, non-focal actor sex, and
non-focal actor race on effectiveness of responses to helping situation 1. The four-way
interaction was not statistically significant, F(1,203) = 1.19, MSE = .87, ns. Thus, the interaction
of respondent sex, respondent race, non-focal actor sex, and non-focal actor race did not
significantly impact how effectively respondents answered helping item 1. Again, the power for
finding an effect was low. What was statistically significant though was the three way interaction
between respondent sex, non-focal actor sex, and non-focal actor race (Black, White), F(1,203) =
4.34, MSE = .87, p < .05. Separate ANOVAs were done by non-focal actor race. With a Black
non-focal actor, men and women did not score differently depending on whether the Black actor
was male or female, F(1,110) = .13, MSE = .88, ns. With a White non-focal actor, however, men

98

and women scored differently on helping item 1 depending on the gender of the actor, F(1,101) =
4.39, MSE = .82, p < .05. Male respondents (M = .74, SD = .90) and female respondents (M =
.69, SD = .97) got equal scores on average on helping item 1 when the White non-focal actor was
female. F(1,51) = .04, MSE = .88, ns, but when the White non-focal actor was male, female
respondents (M = .96, SD = .87) scored higher on the item than male respondents (M = .27, SD =
.87), F(1,50) = 8.18, MSE = .76, p < .01. These differences are displayed in Figure 20 below.

1.20

Male Respondents
.96
(.87)

1.00

.69 .69
(.88) (.97)

.80
Helping Situation 1 Score

Female Respondents
.85
.77 (.86)
(.82)

.47
(1.11)

.60

.40

.68
(.91)

.38
(.82)

.20

.00
Male White
Non-focal
Actor

Female White
Non-Focal
Actor

Male Black
Non-focal
Actor

Female Black
Non-focal
Actor

Figure 20. Effects of non-focal actor race and gender on helping situation 1 score for male
and female respondents.
Standard deviations are below the group means in parentheses.

99

To try to find a substantive explanation for women’s advantage on helping item 1 relative
to men when the non-focal actor was a White male, cross-tabs with a chi-square test of
independence were calculated on their ML and LL responses (to examine how the frequencies
with which different options were selected as most or least likely were affected by respondent
gender). Respondent gender did not have a statistically significant effect on the frequency with
which various response options were selected as most likely,
likely,

2

(5, N = 52) = 8.03, ns, or least

2

(3, N = 52) = 4.25, ns. Actual frequencies of responses were not further examined.

H5a, scoring method 1. Hypothesis 5a predicted that in a helping situation where the
work ethic stereotype is activated, the lowest level of help will be offered by White respondents
to Black non-focal actors as compared to other race-based combinations of respondents with
targets. Scoring method 1 was applied to test this hypothesis at the level of the helping
dimension. Respondent race (Asian, Black, White) and non-focal actor race (Asian, Black,
White) were entered as independent variables into an ANOVA. The main effect for non-focal
actor race was significant, F(2,313) = 3.41, MSE = 30.56, p < .05. There was only a marginally
significant interaction of respondent and non-focal actor race on helping score, F(4,313) = 2.01,
MSE = 30.56, p = .094. Thus, Hypothesis 5a was not supported using scoring method 1.
H5a, scoring method 2. Scoring method 2 was applied next in order to evaluate
Hypothesis 5a at the level of the response options associated with helping situation 2. A
MANOVA was calculated with race of respondent and race of non-focal actor as the independent
variables. The interaction effect on likelihood ratings given to the five response options as a set
was not statistically significant, Wilks’ lambda = .92, F(20, 1025.79) = 1.33, ns. Thus,
Hypothesis 5a was also not supported at the level of these individual response options. The main
effect for non-focal actor race was significant though, Wilks’ lambda = .96, F(10,618) = 1.97, p
100

< .05, but a follow up ANOVA with post hoc Tukey tests on response option 1(for which the
main effect of non-focal actor race had been significant) did not show significant pairwise
differences between likelihood ratings by non-focal race.
H5a, scoring method 3. Next, scoring method 3 was applied to examine the effects of
the interaction of respondent race and non-focal actor race on effectiveness of responses to
helping situation 2. The main effect of respondent race and the interaction of respondent with
non-focal actor race were statistically significant, F(2,313) = 5.35, MSE = .34, p < .01, and
F(4,313) = 2.91, MSE = .34, p < .05, respectively. To break down the interaction separate
ANOVAs were done by respondent race. Helping item 2 scores for Asian respondents and White
respondents did not differ depending on the race of the non-focal actor, F(2,35) = 1.05, MSE =
.36, ns, and F(2,250) = .16, MSE = .32, ns, respectively. Black respondents, on the other hand,
answered the item less effectively (again, as defined by the test developer) when the non-focal
actor was Asian (M = .00, SD = .63) than when the non-focal actor was Black (M = .77, SD =
.73), F(2,28) = 4.32, MSE = .42, p < .05. (And the difference between the Black and White nonfocal actor conditions was not significant as indicated by a post hoc Tukey test.) These results
are shown in Figure 21 below.

101

1

Asian coworker

0.9
0.8
0.7

Helping Situation 2 Score

0.6
0.5
0.4

.69
(.60)

.66
(.60)

Black coworker

.77
(.73)

.71
(.57)

White coworker

.68
(.52)

.45
(.69)
.36
(.50)
.29
(.49)

0.3
0.2
.00
(.63)

0.1
0
Asian

White

Black

Test Taker Race

Figure 21. Effects of non-focal actor race on helping situation 2 score for different groups of
respondents.
Standard deviations are below the group means in parentheses.

To try to find a substantive explanation for the significant effect of non-focal actor race
on Black respondents’ helping situation 2 scores, cross-tabs with a chi-square test of
independence were calculated on their ML and LL responses. Non-focal actor race (Asian,
Black) did not have a statistically significant effect on the frequency with which various response
options were selected as most likely,

2

(4, N = 24) = 3.39, ns (Fisher’s exact test likewise did not

support rejecting the null hypothesis of no effect for non-focal actor race). The effect on least
102

likely responses, however, was significant,

2

(4, N = 24) = 11.12, p < .05 (Fisher’s exact test

likewise supported rejecting the null hypothesis of no effect). Therefore, actual frequencies of
responses could be further examined. Table 15 shows that a greater proportion of Black
respondents in the higher scoring Black non-focal actor group than in the lower scoring Asian
non-focal actor group tended to pick the ―correct‖ response option (3; Tell your coworker you
think he/she is taking advantage of the system and you won't switch) as their least likely
response to helping item 2. For the lower scoring group, the biggest distracter from the correct
response appeared to be response option 1 (Agree to switch with your coworker so that he/she
can take the training course; scored as a ―0‖). Thinking in terms of how judges rated each of
these options on level of helping, the correct least likely response option—3—was the least
helpful one (M = 1.33 on a 3-point scale, SD = .66), while option 1 was one of the most helpful
ones (M = 2.70 on a 3-point scale, SD = .65). While members of the higher scoring group tended
to indicate that acting in an unhelpful manner was the least likely thing that they would do if
asked to switch shifts with a coworker, members of the lower scoring group were not as good at
recognizing and/or showing that this would be the worst way to respond to the situation and were
more likely to indicate instead that one of the more helpful options was something they would be
least likely to do in the situation. The implication based on this analysis is that for helping
scenario 2, Black respondents were more willing to be helpful to a non-focal actor of the same
race than to an Asian non-focal actor. It is interesting to note, though, that this effect did not
emerge in the above reviewed analyses using likelihood ratings assigned to helping response
options as depending variables (H5a, scoring methods 1 and 2).

103

Table 15. Observed frequencies of Black respondents' least likely responses on helping
situation 2
Helping Situation 2 Response Options

Asian Non-focal Actor
Black Non-focal Actor

1
6
54.5%
1
7.7%

2
1
9.1%
0
0.0%

a

3
1
9.1%
7
53.8%

4
1
9.1%
4
30.8%

5
2
18.2%
1
7.7%

Row Total
11
100%
13
100%

a

Note . Considered effective response.

H5b, scoring method 1. To test Hypothesis 5b, that in a helping situation where the
work ethic stereotype is activated, the highest level of help will be offered to Asian non-focal
actors, an ANOVA with non-focal actor’s race as the independent variable was computed. The
ANOVA showed no statistically significant effect of non-focal actor’s race on respondent
willingness to help, F(2,344) = 1.45, MSE = 31.69, ns. Hypotheses 5b was therefore not
supported at the level of the helping dimension (i.e., using scoring method 1).
H5b, scoring method 2. Next, scoring method 2 was applied to calculate a MANOVA
on the response options associated with helping situation 2. Hypothesis 5b was also not
supported at the level of these individual response options as the main effect for non-focal actor
race on likelihood ratings respondents assigned to the two response options was not significant,
Wilks’ lambda = .97, F(10,680) = .92, ns.
H5b, scoring method 3. Next, scoring method 3 was applied to examine the effects of
non-focal actor race on effectiveness of responses to helping situation 2. There was no
significant main effect of non-focal actor race on helping item 2 score, F(2,344) = .43, MSE =
.36, ns. This lack of a main effect on the second helping item is consistent with earlier analyses

104

using this same scoring method 3, which instead of a main effect show non-focal actor race to
interact with respondent race in determining how effectively respondents answer the item.
Next, results pertaining to Hypotheses 6 through 16 (about conflict) are reviewed. Before
proceeding, results for Hypotheses 1 through 5 (about helping) are summarized in Table 16
below. Also, the results of the analyses pertaining to response effectiveness are summarized in
Table 17.

105

Table 16. Summary of findings for helping situations
Hypothesis
Hypothesis 1: Respondent race will moderate the effects of the race mismatch
between the respondent and the non-focal actor on helping, such that the effects
of race mismatch will be more negative for White respondents viewing a
minority non-focal actor than for minority respondents viewing a White nonfocal actor.
Hypothesis 2: There will be a main effect for respondent sex on helping, such
that female respondents will offer more help to non-focal actors than will male
respondents.

Level of Support
Scoring method 1
Scoring method 2

No support

No support

No support

Partial support (HS1 options
3 and 6 (marginal); HS2
options 5 and 2 (marginal))

Hypothesis 3: Respondent sex will moderate the effects of non-focal actor sex
on helping, such that a female non-focal actor will receive more help than a
male non-focal actor when the respondent is male. Female and male non-focal
actors will receive equal amounts of help when the respondent is female.

Supported

No support

Hypothesis 4: In a helping situation where the competence stereotype is
activated, the lowest level of help will be offered by White male respondents to
male Black non-focal actors as compared to other race and sex-based
combinations of respondents with targets.

Could not test

No support

No support

No support

No support

No support

Hypothesis 5a: In a helping situation where the work ethic stereotype is
activated, the lowest level of help will be offered by White respondents to
Black non-focal actors as compared to other race-based combinations of
respondents with targets.
Hypothesis 5b: In a helping situation where the work ethic stereotype is
activated, the highest level of help will be offered to Asian non-focal actors.
Note. HS1 = Helping situation 1. HS2 = Helping situation 2.

106

Table 17. Effects of demographic cues on performance on helping items
Test
Test 1: Does respondent race moderate the effects of
the race mismatch between the respondent and the
non-focal actor on scores?
Test 2: Is there a main effect for respondent sex on
scores?
Test 3: Does respondent sex moderate the effects of
non-focal actor sex on scores?
Test 4: Is there a four-way interaction of respondent
sex, dichotomized respondent race (minority, White),
non-focal actor sex, and non-focal actor race (Black,
White) on helping situation 1 score?

Finding
DV: Helping situation 1 score
No; marginal negative effect of race
mismatch on both minority and White
respondents

DV: Helping situation 2 score
Yes; more negative effects of
race mismatch on minority
than on White respondents

No

No

No

No

No; three-way interaction of respondent
sex, non-focal actor sex, and non-focal
actor race was significant

--

Test 5a: Is there an interaction of respondent race
(Asian, Black, White) and non-focal actor race
(Asian, Black, White) on helping situation 2 score?
Test 5b: Does non-focal actor race (Asian, Black,
White) affect helping situation 2 score?
Note. Test numbers correspond roughly to hypothesis tests in Table 16.

107

--

Yes; Black respondents scored
lower when the non-focal actor
was Asian as opposed to Black

--

No

Tests of Hypotheses Pertaining to Conflict Situations
Hypotheses 6a through 16 could not be tested using scoring method 1 due to the lack of
internal consistency of response options pertaining to both of the conflict situations. Before
discussing hypotheses pertaining to conflict, I include Tables 18 and 19 below to provide an
overview of mean likelihood ratings as a function of sex (Table 18) and race (Table 19) of the
respondent and non-focal actor.

108

Table 18. Likelihood ratings for conflict response options as a function of respondent's and non-focal actor's sex
Female Respondent
Male Respondent
Female nonfocal actor
(N = 96)
Response Option
CS1 Op1: Tell the coworker that you will continue to watch for unglued
cartons and get back to him/her.
CS1 Op2: Assume that the cartons are being glued properly because your
coworker hasn't noticed the problem.
CS1 Op3: Insist that there must be a problem your coworker hasn't
noticed since unglued cartons are arriving at the warehouse.
CS1 Op4: Ask another team member to discuss the problem with your
coworker.
CS1 Op5: Let your team leader know that your coworker is allowing
unglued cartons to get to the warehouse.
CS2 Op1: Ask your coworker what you could do to get along better.
CS2 Op2: Ask the team leader for suggestions about how to work with
your coworker.
CS2 Op3: Try to avoid your coworker whenever possible because you
can't change him/her.
CS2 Op4: Tell your coworker that he/she should act like a team member
and try to work together.
CS2 Op5: Ask the team leader to meet with you and your coworker to
work out your differences.
CS2 Op6: Ask other team members for suggestions on how to work with
your coworker.
Note. CS1 = Conflict situation 1. CS2 = Conflict situation 2.

109

Male nonfocal actor
(N = 93)

Female nonfocal actor
(N = 95)

Male nonfocal actor
(N = 90)

M

SD

M

SD

M

SD

M

SD

4.48

.74

4.14

.89

4.43

.61

4.09

.92

2.55

1.02

2.46

1.06

2.44

1.08

2.60

1.06

3.36

1.09

3.67

1.00

3.47

1.06

3.42

1.04

2.81

1.17

2.66

1.17

2.87

1.04

2.74

1.13

3.15

1.26

3.03

1.17

3.06

1.15

2.87

1.19

4.15

.98

3.87

1.19

3.71

1.20

3.97

.97

3.95

.99

3.76

1.07

3.64

1.12

3.51

1.11

2.36

1.05

2.60

1.15

2.59

1.09

2.34

1.17

3.52

1.10

3.53

1.10

3.47

1.14

3.56

1.03

3.48

1.26

3.19

1.33

3.29

1.30

3.10

1.25

3.58

1.18

3.70

1.17

3.67

1.03

3.82

1.01

Table 19. Likelihood ratings for conflict response options as a function of respondent's and non-focal actor's race
Asian Respondent
Black Respondent
White Respondent
Asian nonfocal actor
(N = 17)
M

SD

Black nonfocal actor
(N = 12)

White nonfocal actor
(N = 12)

Asian nonfocal actor
(N = 11)

M

SD

M

SD

M

SD

Black nonfocal actor
(N = 14)
M

SD

White nonfocal actor
(N = 10)
M

SD

Asian nonfocal actor
(N = 94)
M

SD

Black nonfocal actor
(N = 89)
M

SD

White nonfocal actor
(N = 88)
M

SD

CS1.1 4.24

.66

3.92

1.31

3.75

1.22

4.55

.52

4.64

.50

4.20

.63

4.34

.84

4.30

.74

4.32

.75

CS1.2 2.76

1.15

2.17

1.27

3.25

1.29

2.36

1.12

2.36

1.22

2.90

1.10

2.49

1.03

2.34

.90

2.65

1.08

CS1.3 3.82

.73

3.50

1.24

3.67

.89

3.27

.90

4.07

.73

4.10

.74

3.45

1.03

3.43

1.02

3.19

1.15

CS1.4 3.12

1.17

3.00

1.28

3.17

1.34

3.00

.89

2.64

1.22

2.80

1.48

2.70

1.08

2.82

1.13

2.63

1.11

CS1.5 3.00

1.37

2.92

1.51

2.92

1.31

3.45

1.13

3.14

1.23

3.10

1.52

3.02

1.22

3.18

1.08

2.76

1.15

CS2.1 3.88

1.22

3.50

1.17

4.08

1.08

3.55

1.44

3.93

.92

3.70

1.25

3.89

1.18

4.12

.91

3.88

1.12

CS2.2 3.47

1.23

3.83

1.19

3.50

1.57

3.91

.83

3.79

1.12

3.60

1.17

3.74

1.07

3.87

1.08

3.67

.97

CS2.3 2.24

.75

2.92

1.44

2.75

1.22

2.73

1.19

1.71

.83

2.20

1.40

2.38

1.09

2.48

1.06

2.57

1.08

CS2.4 3.71

.69

3.08

1.38

3.75

.87

3.73

1.42

3.29

1.27

4.40

.52

3.38

1.13

3.56

1.09

3.42

1.08

CS2.5 3.29

1.10

2.50

1.45

3.50

1.09

4.09

1.22

3.64

1.39

3.00

1.33

3.29

1.23

3.38

1.23

2.97

1.37

CS2.6 3.94 .43
4.00 1.04 4.17 .39
3.64 1.21
Note. CS1 = Conflict situation 1. CS2 = Conflict situation 2.

3.00

1.41

3.00

1.63

3.72

1.08

3.61

1.16

3.75

1.02

110

H6, scoring method 2. Method 2 was used to evaluate Hypothesis 6, that in responses to
conflict situations there will be a main effect for respondent-non-focal actor race match on
assertiveness (6a) and cooperativeness (6b), in an exploratory manner (by using unweighted
likelihood ratings). A MANOVA was performed to examine the impact of race match on
likelihood ratings for the 11 response options associated with the 2 conflict items. There were no
significant differences in likelihood ratings based on whether the respondent’s and non-focal
actor’s race matched or not, Wilks’ lambda = .97, F(11,285) = .74, ns. This likely indicates that
Hypothesis 6 would not have received support if the more appropriate scoring method 1 (with
weighted scores) was used.
H6, scoring method 3. Method 3 was then applied to test for main effects of respondentnon-focal actor race match on effectiveness (as defined by the test developer) of responses to
conflict situations 1 and 2. There was no significant main effect for race match on neither
conflict item 1 score nor conflict item 2 score, F(1,318) = .39, MSE = 1.01, ns, and F(1,316) =
.18, MSE = .70, ns, respectively.
H7, scoring method 2. To evaluate Hypothesis 7, that there will be an interaction
between respondent race and respondent-non-focal actor race match on preference for third party
intervention when the goal is reconciliation, in an exploratory manner, a MANOVA was again
used. The 6 response options associated with conflict situation 2 (where the goal is
reconciliation) were entered in as the dependent variables. The analysis showed no statistically
significant interaction of respondent race (minority, White) and race match of respondent and
non-focal actor on likelihood ratings, Wilks’ lambda = .97, F(6,309) = 1.82, p = .096. Thus,
Hypothesis 7 was not supported using scoring method 2.

111

H7, scoring method 3. Scoring method 3 was next applied to test for an interaction of
respondent race and respondent-non-focal actor race match on effectiveness of responses to
conflict situation 2. The ANOVA did not show a statistically significant interaction, F(1,314) =
1.02, MSE = .70, ns.
H8, scoring method 2. Next, scoring method 2 was applied to assess Hypothesis 8, that
there will be an interaction between respondent race and respondent-non-focal actor race match
on preference for third party intervention when the goal is reporting a coworker’s poor
performance. The MANOVA with the 5 response options associated with conflict situation 1
(where the goal of intervention is reporting a coworker’s poor performance) entered in as
dependent variables showed no statistically significant interaction of respondent race (minority,
White) and race match of respondent and non-focal actor on likelihood ratings, Wilks’ lambda =
.97, F(5,312) = 1.72, ns. It is therefore likely that Hypothesis 8 would not have been supported if
it were possible to test it directly (using method 1) as had been planned.
H8, scoring method 3. Scoring method 3 was applied to test for an interaction of
respondent race and respondent-non-focal actor race match on effectiveness of responses to
conflict situation 1. The ANOVA did not show a statistically significant interaction, F(1,316) =
1.13, MSE = 1.01, ns.
H9, scoring method 2. Next, Hypothesis 9, that female respondents will respond to
conflict with lower assertiveness (9a) and higher cooperativeness (9b) than will male
respondents, was explored. A MANOVA was performed to examine the impact of respondent
gender on likelihood ratings for the 11 response options associated with the 2 conflict items.
There were no significant differences in likelihood ratings based on respondent gender, Wilks’

112

lambda = .97, F(11,309) = .77, ns. This may indicate that Hypothesis 9 would not have received
support if it were possible to test it using scoring method 1 as planned.
H9, scoring method 3. Next, the main effect of gender on effectiveness of responses to
conflict situations 1 and 2 was evaluated using scoring method 3. There was no significant main
effect of gender on either item, F(1,343) = .48, MSE = 1.00, ns, and F(1,344) = .03, MSE = .72,
ns, respectively.
H10, scoring method 2. Hypothesis 10, that respondent gender will moderate the effects
of gender match between the respondent and the non-focal actor on assertiveness (10a) and
cooperativeness (10b) when responding to conflict was addressed next. A MANOVA showed a
statistically significant effect of the gender by gender match interaction on likelihood ratings for
the 11 response options associated with the 2 conflict items, Wilks’ lambda = .93, F(11,307) =
2.05, p < .05. ANOVAs were used to further examine likelihood ratings for the two response
options that appeared to be influenced by the gender by gender match interaction. An ANOVA
on option 1 of conflict situation 1 (Tell your coworker you will continue to watch for unglued
cartons and get back to him/her) showed a statistically significant effect of the interaction on
likelihood ratings, F(1, 341) = 13.55, MSE = .63, p < .001. Separate ANOVAs by gender of
respondent showed that men endorse response option 1 significantly less strongly when the
coworker is of the same gender (i.e., male), F(1,168) = 7.84, MSE = .59, p < .01, while women
endorse response option 1 significantly more strongly when the coworker is of the same gender
(i.e., female), F(1,173) = 5.85, MSE = .67, p < .05.
Given that a higher willingness to respond to the unglued cartons problem by telling the
coworker that he/she will continue to monitor the cartons situation (and not really solve the
problem) corresponds to a low level of assertiveness according to judges’ ratings (M = 1.88 on a

113

3-pt scale, SD = .91; lower mean than for 3 of the 4 other response options), this result is
consistent with the expectation that gender match will have a positive effect on assertiveness for
male respondents and a negative effect on assertiveness for female respondents. On the flipside,
because a higher willingness to respond to the unglued cartons problem this way corresponds to
a higher level of cooperativeness according to judges’ ratings (M = 2.55 on a 3-pt scale, SD =
.78), this result is consistent with the expectation that gender match will have a negative effect on
cooperativeness for male respondents and a positive effect on cooperativeness for female
respondents. Thus, both male and female respondents are more willing to not be very assertive
(or to be more cooperative) with a coworker over a cartons situation when the coworker is
female. These results with respect to conflict situation 1’s response option 1, then, lend some
amount of support to Hypothesis 10.
Next, an ANOVA on option 4 of conflict situation 1 (Ask another team member to
discuss the problem with your coworker) showed the effect of the interaction of gender and
gender match failing to reach a statistically significant effect on likelihood ratings, F(1,341) =
2.53, MSE = 1.25, p = .113. A closer look at the cell means, however, revealed that they trended
toward the same direction as those for response option 1 discussed above: respondents were
somewhat more likely to endorse response option 4 when the coworker was female as opposed to
male. Yet, the results with respect to conflict situation 1’s response option 4 do not lend support
to Hypothesis 10 in a statistically significant manner.
H10, scoring method 3. Next, scoring method 3 was applied to examine whether
respondent gender moderated the effects of gender match between the respondent and the nonfocal actor on effectiveness of responses to the conflict items. The interaction of respondent
gender with gender match on conflict item 1 score was not statistically significant, F(1,341) =

114

.81, MSE = 1.00, ns. The effect on conflict item 2 score was likewise not significant, F(1,342) =
2.28, MSE = .72, ns.
H11, scoring method 2. To explore the tenability of Hypothesis 11, that when the goal of
the conflict situation is reconciliation both male and female respondents will prefer third party
intervention more when the non-focal actor is female, a MANOVA with the 6 response options
associated with conflict situation 2 (where the goal is reconciliation) entered in as dependent
variables was calculated. The gender by gender match interaction did not have a significant
effect on likelihood ratings on the 6 response options associated with conflict situation 2, Wilks’
lambda = .98, F(6,337) = .90, ns. This may indicate that Hypothesis 11 would not have received
support if it were possible to test it as planned. And as already reviewed above, there is also no
significant interaction of respondent gender with gender match on effectiveness of responses to
conflict situation 2.
H12, scoring method 2. A MANOVA was likewise used to explore Hypothesis 12, that
when the goal of the conflict situation is reporting a coworker’s poor performance, both male and
female respondents will show less preference for third party intervention when the non-focal
actor is female. There was a statistically significant effect of the gender by gender match
interaction on likelihood ratings for the 5 response options associated with conflict situation 1
(where the goal of intervention would be to report a coworker’s poor performance), Wilks’
lambda = .94, F(5,337) = 4.07, p = .001. Next, the follow up analysis for option 1 of conflict
situation 1 (Tell your coworker you will continue to watch for unglued cartons and get back to
him/her) matches that of Hypothesis 10 above and the results are exactly the same: men endorse
response option 1 significantly less strongly when the coworker is of the same gender (i.e.,
male), F(1,168) = 7.84, MSE = .59, p < .01, while women endorse response option 1

115

significantly more strongly when the coworker is of the same gender (i.e., female), F(1,173) =
5.85, MSE = .67, p < .05. In the context of the analysis of Hypothesis 12, this suggests that both
male and female respondents are less willing to get a third party to intervene in an unglued
cartons situation when the coworker is female. This result for response option 1 of conflict
situation 1 is consistent with what Hypothesis 12 predicts will happen. As already reviewed
earlier, there is no significant interaction of respondent gender with gender match on
effectiveness of responses to conflict situation 1, so method 3 was not applied to test again for an
interaction on conflict item 1 scores.
H13, scoring method 2. Hypothesis 13 predicted that in conflict situations the lowest
level of assertiveness will be exhibited by female respondents paired with a female non-focal
actor of a different race (13a) and that the highest level of assertiveness will be exhibited by male
respondents paired with a male non-focal actor of the same race (13b), as compared to other sex
and race-based combinations of respondents with targets. To explore these proposed
relationships, gender of respondent, gender of non-focal actor, and race match of respondent
relative to the non-focal actor (match, mismatch) were entered as independent variables into a
MANOVA. The analysis did not show a statistically significant effect of the three way
interaction on the likelihood ratings respondents gave to the 11 response options associated with
the two conflict situations, Wilks’ lambda = .96, F(11,279) = .97, ns. The main effect for nonfocal actor gender was significant, Wilks’ lambda = .93, F(11,279) = 2.04, p < .05. This was due
to conflict item 1’s response option 1 being rated differently by respondents depending on
whether the non-focal actor was male or female. This finding was discussed earlier in the context
of Hypothesis 10. Thus, because the three way interaction was not significant, there was no

116

support for Hypotheses 13a or 13b based on an analysis using likelihood ratings for individual
response options.
H14, scoring method 2. Hypothesis 14 proposed that in conflict situations the lowest
level of cooperativeness will be exhibited by male respondents paired with male non-focal actors
of a different race (14a) and that the highest level of cooperativeness will be exhibited by female
respondents paired with a female non-focal actor of the same race (14b) as compared to other sex
and race-based combinations of respondents with targets. An exploration of this hypothesis at the
individual response option level looks similar to that of Hypothesis 13 above. Thus, there was no
support for Hypotheses 14a and 14b based on an analysis using likelihood ratings for individual
response options.
H14, scoring method 3. Next, method 3 was applied to examine whether the interaction
of gender of respondent, gender of non-focal actor, and race match of respondent relative to the
non-focal actor (match, mismatch) affected effectiveness of responses to conflict situations 1 and
2. The 3-way interaction on conflict item 1 score was not significant, F(1,312) = 2.18, MSE =
1.01, ns, and neither was the 3-way interaction on conflict item 2 score, F(1,310) = .04, MSE =
.70, ns. None of the lower order interactions or main effects were statistically significant either.
H15, scoring method 2. Hypothesis 15 proposed that in a conflict situation where the
goal of involving a third party is reconciliation, the highest preference for third party intervention
will be exhibited by collectivist (Asian, Black) respondents paired with a female non-focal actor
of the same race. To explore this proposed relationship a MANOVA was calculated. The
independent variables entered into the analysis were dichotomized respondent race (minority,
White), non-focal actor sex, and race match between respondent and non-focal actor (match,
mismatch). The proposed three way interaction on the six response options associated with

117

conflict item 2 (where the goal of third party intervention would be reconciliation) was not
statistically significant, Wilks’ lambda = .99, F(6,305) = .46, ns. None of the lower order
interactions or main effects were significant either. These results at the level of the response
options do not lend support to Hypothesis 15.
H15, scoring method 3. Next, scoring method 3 was used to check for a three way
interaction of dichotomized respondent race (minority, White), non-focal actor sex, and race
match between respondent and non-focal actor (match, mismatch) on effectiveness of responses
to conflict item 2. The three way interaction on conflict item 2 score was marginally significant,
F(1,310) = 3.66, MSE = .69, p = .057. A lower order interaction of respondent race with nonfocal actor sex was also significant, F(1,310) = 6.74, MSE = .69, p = .01. The main effect for
non-focal actor sex was significant as well, F(1,310) = 5.13, MSE = .69, p < .05. To break down
the significant two-way interaction separate ANOVAs were calculated by dichotomized
respondent race. White respondents scored the same on the item regardless of whether the nonfocal actor was female (M = .57, SD = .84) or male (M = .55, SD = .81). Minority (Asian, Black)
respondents who saw a female non-focal actor scored higher on conflict item 2 (M = .74, SD =
.76) than minority respondents who saw a male non-focal actor (M = .26, SD = .92), F(1,71) =
5.95, MSE = .71, p < .05. To try to understand this effect of non-focal actor gender on minority
respondents’ conflict situation 2 scores, cross-tabs with a chi-square test of independence were
calculated on their ML and LL responses. Non-focal actor gender did not have a statistically
significant effect on the frequency with which various response options were selected as most
likely,

2

(5, N = 73) = 8.76, ns, or least likely,

2

(5, N = 73) = 9.74, ns. Therefore, actual

frequencies of responses were not examined.

118

H16, scoring method 2. Next, Hypothesis 16 was explored using scoring method 2. It
stated that in a conflict situation where the goal of involving a third party is reporting a
coworker’s poor performance, the highest preference for third party intervention will be
exhibited by collectivist (Asian, Black) respondents paired with a male non-focal actor of a
different race. A MANOVA was calculated using the response options associated with conflict
situation 1 (where the goal of third party intervention would be reporting a coworker’s poor
performance). The independent variables entered into the analysis were dichotomized respondent
race (minority, White), non-focal actor sex, and race match between respondent and non-focal
actor (match, mismatch). The proposed three way interaction on the five response options
associated with conflict item 1 was not statistically significant, Wilks’ lambda = .99, F(5,308) =
.604, ns. The main effects for dichotomized respondent race and for non-focal actor sex were
significant, Wilks’ lambda = .96, F(5,308) = 2.66, p < .05, and Wilks’ lambda = .96, F(5,308) =
2.45, p < .05, respectively. As mentioned several times before, the main effect for non-focal
actor sex is attributable to conflict item 1’s response option 1 being rated differently by
respondents depending on whether the non-focal actor was male or female. The main effect for
dichotomized respondent race had to do with minorities giving higher likelihood ratings on
average (M = 3.79, SD = .85) to response option 3 (Insist that there must be a problem your
coworker hasn't noticed since unglued cartons are arriving at the warehouse) than White
respondents (M = 3.39, SD = 1.07), F(1,321) = 8.34, MSE = 1.06, p < .01. These results at the
level of the response options do not lend support to Hypothesis 16.
H16, scoring method 3. Scoring method 3 was used to check for a three way interaction
of dichotomized respondent race (minority, White), non-focal actor sex, and race match between
respondent and non-focal actor (match, mismatch) on effectiveness of responses to conflict item

119

1. The three way interaction on conflict item 1 score was not statistically significant, F(1,312) =
.07, MSE = .99, ns. The main effect for non-focal actor sex was significant, F(1,312) = 4.08,
MSE = .99, p < .05, and so was the two-way interaction of dichotomized respondent race and
non-focal actor sex, F(1,312) = 6.85, MSE = .99, p < .01. The nature of this two-way interaction
was such that White respondents’ scores were not impacted by whether the non-focal actor was
female (M = 1.19, SD = 1.02) or male (M = 1.28, SD = .95), F(1,249) = .54, MSE = .97, ns,
whereas minority (Asian, Black) respondents scored higher on conflict item 1 when the nonfocal actor was female (M = 1.42, SD = .83) than when that actor was male (M = .76, SD = 1.18),
F(1,70) = 7.58, MSE = 1.02, p < .01. These results are consistent with those reported above
(H15, scoring method 3) for conflict item 2. To try to understand this effect of non-focal actor
gender on minority respondents’ conflict situation 1 scores, cross-tabs with a chi-square test of
independence were calculated on their ML and LL responses. Non-focal actor gender did not
have a statistically significant effect on the frequency with which various response options were
selected as most likely,
marginally significant,

2

(4, N = 72) = 3.70, ns, and the effect on least likely responses was only

2

(4, N = 72) = 9.37, p = .053. Therefore, actual frequencies of responses

were not examined.
Next, results pertaining to Hypotheses 17 and 18 (about applicant reactions) are
reviewed. Before proceeding, results for Hypotheses 6 through 16 (about conflict) are
summarized in Table 20 below. Also, the results of the analyses pertaining to response
effectiveness are summarized in Table 21.

120

Table 20. Summary of findings for conflict situations
Level of
a
Support

Hypothesis
Hypothesis 6a: In responses to conflict situations respondents will endorse a
more assertive response when the respondent and the non-focal actor are of the
same (vs different) race.
Hypothesis 6b: In responses to conflict situations respondents will endorse a
more cooperative response when the respondent and the non-focal actor are of
the same (vs different) race.
Hypothesis 7: When the goal is reconciliation respondents belonging to
collectivist groups (Asians, Blacks) will show a preference for third party
intervention when paired with non-focal actors of the same (vs different) race as
them. Race match between respondents and non-focal actors will not influence
preferences for third party intervention among individualist respondents
(Whites).
Hypothesis 8: When the goal of third party intervention is reporting a
coworker’s poor performance respondents belonging to collectivist groups
(Asians, Blacks) will show a preference for third party intervention when paired
with non-focal actors of a different (vs same) race as them. Race match between
respondents and non-focal actors will not influence preferences for third party
intervention among individualist respondents (Whites).
Hypothesis 9a: Female respondents will respond to conflict with lower
assertiveness than male respondents.
Hypothesis 9b: Female respondents will respond to conflict with higher
cooperativeness than male respondents.
Hypothesis 10a: For male respondents, gender match with the non-focal actor
will have a positive effect on assertiveness. For female respondents, gender
match will have a negative effect on assertiveness.
Hypothesis 10b: For male respondents, gender match with the non-focal actor
will have a negative effect on cooperativeness. For female respondents, gender
match will have a positive effect on cooperativeness.
Hypothesis 11: When the goal of the conflict situation is reconciliation: a.
Female respondents will show a preference for third party intervention when
paired with female (vs male) non-focal actors; b. Male respondents will show a
preference for third party intervention when paired with female (vs male) nonfocal actors.
Hypothesis 12: When the goal of the conflict situation is reporting a coworker’s
poor performance: a. Female respondents will show less of a preference for
third party intervention when paired with female (vs male) non-focal actors; b.
Male respondents will show less of a preference for third party intervention
when paired with female (vs male) non-focal actors.

121

No support

No support

No support

No support

No support
No support
Minimal (CS1
option 4)
Minimal (CS1
option 4)

No support

Minimal (CS1
option 1)

Table 20 (cont’d)
Hypothesis 13a: In conflict situations the lowest level of assertiveness will be
exhibited by female respondents paired with a female non-focal actor of a
different race as compared to other sex and race-based combinations of
respondents with targets.
Hypothesis 13b: In conflict situations the highest level of assertiveness will be
exhibited by male respondents paired with a male non-focal actor of the same
race as compared to other sex and race-based combinations of respondents with
targets.
Hypothesis 14a: In conflict situations the lowest level of cooperativeness will
be exhibited by male respondents paired with male non-focal actors of a
different race as compared to other sex and race-based combinations of
respondents with targets.
Hypothesis 14b: In conflict situations the highest level of cooperativeness will
be exhibited by female respondents paired with a female non-focal actor of the
same race as compared to other sex and race-based combinations of respondents
with targets.
Hypothesis 15: In a conflict situation where the goal of involving a third party is
reconciliation, the highest preference for third party intervention will be
exhibited by collectivist (Asian, Black) respondents paired with a female nonfocal actor of the same race.
Hypothesis 16: In a conflict situation where the goal of involving a third party is
reporting a coworker’s poor performance, the highest preference for third party
intervention will be exhibited by collectivist (Asian, Black) respondents paired
with a male non-focal actor of a different race.
a

No support

No support

No support

No support

No support

No support

Note . Hypotheses were tested using scoring method 2; scoring method 1 could not be applied.
CS1 = Conflict situation 1.

122

Table 21. Effects of demographic cues on performance on conflict items
Finding
DV: Conflict situation 1
score

DV: Conflict situation 2
score

No

No

No

No

Test 9: Is there a main effect for respondent gender on scores?

No

No

Test 10/11/12: Is there an interaction of respondent gender and gender match
between respondent and non-focal actor on scores?

No

No

Test 13/14: Is there a three-way interaction of respondent gender, non-focal
actor gender, and race match between respondent and non-focal actor on
scores?

No

No

Test 15/16: Is there a three way interaction of dichotomized respondent race
(minority, White), non-focal actor gender, and race match between
respondent and non-focal actor on scores?

No; two-way interaction
of dichotomized
respondent race and nonfocal actor gender was
significant

No; two-way interaction
of dichotomized
respondent race and nonfocal actor gender was
significant

Test
Test 6: Does race match between the respondent and the non-focal actor
affect scores?
Test 7/8: Is there an interaction of respondent race and race match between
respondent and non-focal actor on scores?

Note. Test numbers correspond roughly to hypothesis tests in Table 20.

123

Correlations with Applicant Reactions
Looking at the correlations between dichotomized respondent race and applicant
reactions shown in Table 10, minorities apparently perceived higher opportunity to perform on
the test (r = -.13) and higher job relatedness of the test (both predictive and content-based; r = .14 and r = -.13, respectively). T-tests (on the respondents who passed all four manipulation
checks) confirmed that as a group, Asian and Black respondents reported higher opportunity to
perform (M = 3.18, SD = .86) than White respondents (M = 2.85, SD = .95), t(276) = 2.48, p <
.05, as well as higher job relatedness-predictive (M = 2.31 and SD = 1.09 compared to M = 1.91
and SD = .84), t(86.65) = 2.74, p < .01, and higher job relatedness-content (M = 2.36 and SD =
.95 compared to M = 2.01 and SD = .88), t(276) = 2.74, p < .01. However, when ANOVAs were
used to examine differences by respondent race (Asian, Black, White), post hoc Tukey tests
showed none of the pairwise differences to be statistically significant.
Looking again at the correlations in Table 10, respondent gender was correlated with
perceived compatibility with the organization (r = .17), indicating more positive reactions from
male respondents. T-tests (on the respondents who passed all four manipulation checks) likewise
showed that male respondents reported higher compatibility with the organization (M = 3.32, SD
= .95) than female respondents (M = 2.93, SD = .84), t(296) = -3.80, p < .001. Additionally, the
difference between men and women on attraction toward the organization was marginally
significant, t(296) = -1.96, p = .051, with men again reacting more positively (M = 3.71, SD =
.85) than women (M = 3.53, SD = .78).

124

Tests of Hypotheses Pertaining to Applicant Reactions
Before discussing hypotheses pertaining to applicant reactions, I include Tables 22 and
23 below to provide an overview of reactions as a function of sex (Table 22) and race (Table 23)
of the respondent and non-focal actor.

125

Table 22. Reactions as a function of respondent's and non-focal actor's sex
Female Respondent
Male Respondent
Female
Male
Female
non-focal
non-focal
non-focal
Male nonactor
actor
Total
actor
focal actor
Total
(N = 96)
(N = 93) (N = 189)
(N = 95)
(N = 90)
(N = 185)
Measure
M
SD
M SD M SD
M
SD
M
SD
M
SD
Opportunity to Perform
2.94 .98 2.94 .87 2.94 .92 2.93 .98 2.96 .92 2.95 .95
Job Relatedness-Predictive
2.05 1.05 1.99 .83 2.02 .94 2.01 .89 1.95 .89 1.98 .89
Job Relatedness-Content
2.06 .98 2.09 .94 2.07 .96 2.23 .97 2.22 .95 2.22 .96
Attraction
3.59 .77 3.49 .81 3.54 .79 3.67 .92 3.71 .80 3.69 .86
Compatibility
3.01 .82 2.99 .84 3.00 .83 3.27 .99 3.33 .90 3.30 .95

126

Table 23. Reactions as a function of respondent's and non-focal actor's race
Asian Respondent
Black Respondent
Asian
White
non-focal Black non- non-focal Asian non- Black non- White nonactor
focal actor
actor
focal actor focal actor focal actor
(N = 17)
(N = 12)
(N = 12)
(N = 11)
(N = 14)
(N = 10)
Measure
M SD
M
SD
M SD
M
SD
M
SD
M
SD
Opportunity
to Perform
Job Related.Predictive
Job Related.Content
Attraction
Compatibility

White Respondent
Asian
Black
White
non-focal non-focal non-focal
actor
actor
actor
(N = 94)
(N = 89)
(N = 88)
M SD
M SD M SD

3.09

.78

3.42

.62

3.06

.91

3.18

.73

3.46

.81

2.80

1.10

2.82

.96

2.99

.98

2.84

.93

2.26

.97

2.13

1.07

2.25

.94

2.82

1.12

1.96

1.03

1.95

1.04

2.02

.96

1.93

.81

1.86

.84

2.47

.86

1.83

.98

2.38

.77

2.23

.88

2.61

1.15

2.25

.82

2.01

.91

2.18

.92

1.95

.89

3.66
3.00

.68
.91

3.07
2.77

.60
.64

3.79
3.00

.69
.76

3.62
3.11

.84
.72

3.92
3.43

.79
1.06

3.52
2.95

.45
.62

3.58
3.16

.87
.93

3.61
3.15

.80
.87

3.66
3.18

.86
.88

127

H17. Hypothesis 17 stated that minority respondents will react more positively when the
SJT features Black or Asian non-focal actors than when the SJT features White non-focal actors.
Because the dependent variables, attraction toward the organization and perceived compatibility
with the organization, were both significantly correlated with performance on the test (r = .23
and r = .18, respectively; see Table 10), total SJT score was controlled for in the analyses. For
consistency, SJT score was likewise controlled for in the analyses with opportunity to perform
and job relatedness of the test. Further, the analysis excluded individuals who failed one or more
of the manipulation checks (i.e., identifying the race and gender of the actors). An ANCOVA
showed a statistically significant effect of non-focal actor race on minority respondents’
perceived opportunity to perform, F(1,61) = 4.51, MSE = .71, p < .05. Minority respondents
perceived better opportunity to perform when the non-focal actor was a minority (Asian, Black)
(M = 3.31, SD = .75) as opposed to White (M = 2.79, SD = 1.04). ANCOVAs showed no
statistically significant effects of non-focal actor race on minority respondents’ perceived
predictive-type job relatedness of the test, F(1,61) = .27, MSE = 1.10, ns, content-type job
relatedness of the test, F(1,61) = .07, MSE = .93, ns, attraction toward the organization, F(1,61) =
.03, MSE = .53, ns, and on perceived compatibility with the organization, F(1,61) = .27, MSE =
.86, ns. Hypothesis 17 was partially supported.
It should also be noted that ANCOVAs showed no statistically significant effects of nonfocal actor race on White respondents’ perceived opportunity to perform, F(1,211) = .02, MSE =
.87, ns, predictive-type job relatedness of the test, F(1,211) = .80, MSE = .71, ns, content-type
job relatedness of the test, F(1,211) = .35, MSE = .77, ns, attraction toward the organization,
F(1,211) = 1.53, MSE = .66, ns, and on perceived compatibility with the organization, F(1,211) =
.65, MSE = .71, ns.

128

H18. Hypothesis 18 predicted that female respondents will react more positively when
the non-focal actors are male. ANCOVAs showed no statistically significant effects of non-focal
actor gender on women’s perceived opportunity to perform, F(1,143) = .36, MSE = .83, ns,
predictive-type job relatedness of the test, F(1,143) = .85, MSE = .92, ns, content-type job
relatedness of the test, F(1,143) = .46, MSE = .86, ns, attraction toward the organization,
F(1,143) = 2.77, MSE = .60, ns, and on perceived compatibility with the organization, F(1,143) =
.84, MSE = .70, ns. Hypothesis 18 was not supported.
It should also be noted that ANCOVAs likewise showed no statistically significant
effects of non-focal actor gender on men’s perceived opportunity to perform, F(1,149) = 3.49,
MSE = .87, p = .064, predictive-type job relatedness of the test, F(1,149) = .87, MSE = .78, ns,
content-type job relatedness of the test, F(1,149) = .33, MSE = .92, ns, attraction toward the
organization, F(1,149) = .00, MSE = .67, ns, and on perceived compatibility with the
organization, F(1,149) = .07, MSE = .85, ns.
Considering the overall findings with regard to applicant reactions, as a group, minorities
perceived better opportunity to perform when the non-focal actor was a minority (Asian, Black),
and relative to White respondents, they perceived higher job relatedness (content and predictive)
of the test in general. White respondents did not react to the test differently based on the race of
non-focal actors. Female respondents did not react to the test differently based on the gender of
non-focal actors but perceived worse compatibility with the organization in general relative to
male respondents. Female respondents did not differ significantly from male respondents on any
of the other four reactions measures. Before discussing exploratory analyses using overall test
score, results for hypotheses pertaining to applicant reactions are summarized in Table 24 below.

129

Table 24. Summary of findings for applicant reactions
Hypothesis
Hypothesis 17: Minority respondents (Asian, Black) will react more
positively when the SJT features Black or Asian non-focal actors than when
the SJT features White non-focal actors.
Hypothesis 18: Female respondents will react more positively when the nonfocal actors are male.

130

Level of Support
Partial support (minority respondents
perceived better opportunity to
perform when non-focal actor was a
minority)
No support

Exploratory Analyses Using Overall Test Score
As described earlier, respondents’ most and least likely responses to the four SJT items
were scored according to the scoring program originally used by the organization that provided
their SJT for this study. Respondents’ scores ranged from -3 to 8.
The correlations of test score with other variables in the study are presented in Table 10.
SJT scores are positively associated with perceptions of opportunity to perform (r = .15),
attraction to the hiring organization (r = .23), and perceived compatibility with the hiring
organization (r = .18). SJT scores are not related to perceptions of the job relatedness of the test
though.
As a first step to exploring the variables that impact overall SJT scores, a four way
interaction of the demographic (respondent race, respondent gender) and relational demographic
(race match between respondent and non-focal actor, gender match between respondent and nonfocal actor) variables in this study was tested. An ANOVA showed the four way interaction of
these variables on SJT score to not be statistically significant, F(2,250) = .75, MSE = 3.53, ns.
However, a lower order interaction of respondent race by race match between respondent and
non-focal actor, as well as a main effect for race match were statistically significant (p < .05).
The ANOVA results are displayed in Table 25.

131

Table 25. Results of ANOVA for effects of race, gender, race match, gender
match, and their interactions on test scores
SJT Score
2

Partial η
Effect
F
p
.23
.79
.00
Race (R): Asian, Black, White
.01
.90
.00
Gender (G): female, male
6.24
.01
.02
Race match (RM): different race, same race
1.64
.20
.01
Gender match (GM): different gender, same gender
.18
.84
.00
RxG
4.01
.02
.03
R x RM
2.07
.13
.02
R x GM
.00
.99
.00
G x RM
.38
.54
.00
G x GM
.68
.41
.00
RM x GM
.11
.90
.00
R x G x RM
1.54
.22
.01
R x G x GM
.89
.41
.01
R x RM x GM
.12
.72
.00
G x RM x GM
.75
.47
.01
R x G x RM x GM
Note. Total n = 274. Respondents who were not uniquely Asian, Black, or White (n =
27) and respondents who failed one or more manipulation check (n = 76) were
excluded.

To break down the race by race match interaction, separate ANOVAs by race of
respondent were calculated. As can be seen in Figure 22, there was no statistically significant
effect of race match on Asian respondents, F(1,34) = .11, MSE = 4.24, ns, and no statistically
significant effect of race match on White respondents, F(1,221) = .05, MSE = 3.32, ns. There
was, however, a statistically significant effect of race match on Black respondents, F(1,32) =
6.78, MSE = 3.40, p < .05. Black respondents scored higher on the SJT when the non-focal actor
was Black (M = 3.76, SD = 1.79) as opposed to White or Asian (M = 2.12, SD = 1.90).

132

3.76
(1.78)

4

Different-race coworker

3.5

Test Score

3

Same-race coworker

2.72
(1.96) 2.50
(2.15)

2.93 2.99
(1.87) (1.80)
2.12
(1.90)

2.5
2
1.5

1
0.5
0
Asian

White

Black

Test Taker Race

Figure 22. Effects of race match with non-focal actors on test scores for different groups of
respondents.
Standard deviations are below the group means in parentheses.

Consistent with analyses reported earlier (in the context of testing hypotheses pertaining
to helping and conflict) scores for Black respondents did not differ significantly on conflict item
1, t(32) = -2.00, p = .054, on conflict item 2, t(32) = -1.37, ns, and on helping item 1, t(32) = .44, ns. As can be seen in Figure 23, the statistically significant difference was on helping item 2,
t(32) = -2.46, p < .05. Black respondents got a higher score on average on this item (deciding
how to respond to a coworker’s request to switch shifts) when the coworker in the video was
Black (M = .76, SD = .66) as opposed to White or Asian (M = .18, SD = .73).

133

1
Same-race coworker
0.9
0.8

Helping Situation 2 Score

0.7

.76
(.66)

Different-race coworker

.67
.62 (.59)
(.55)

.67
(.59)

0.6
0.5
0.4
0.3

.28
(.57)
.18
(.73)

0.2
0.1
0
Asian

White

Black

Test Taker Race

Figure 23. Effects of race match with the non-focal actor on helping situation 2 scores for
different groups of respondents.
Standard deviations are below the group means in parentheses.

Recall that item scores were calculated based on most likely and least likely responses.
Thus, to come up with a substantive explanation for the effect of race match on Black
respondents’ helping situation 2 scores, cross-tabs with a chi-square test of independence were
calculated on their ML and LL responses to that SJT item. The goal was to examine how the
frequencies with which different response options were selected as most or least likely were
affected by Black respondents’ race match with the non-focal actor in the video. Race match did
not have a statistically significant effect on the frequency with which various response options

134

were selected as most likely,

2

(4, N = 34) = 4.43, ns (Fisher’s exact test likewise did not support

rejecting the null hypothesis of no effect for race match). Race match also did not have a
statistically significant effect on the frequency with which various response options were
selected by Black respondents as least likely,

2

(4, N = 34) = 6.39, ns (Fisher’s exact test

likewise did not support rejecting the null hypothesis of no effect for race match). Substantive
analysis was therefore not justified.
Given the finding that Black respondents performed better on helping item 2 when the
non-focal actor was Black, an analysis of mean differences between Blacks and Whites by race
of the non-focal actor seems pertinent (see Table 26). Black test takers performed about as well
as White test takers on the item in the Black non-focal actor condition, t(94) = .33, ns, but
substantially worse (M = .11, SD = .58) than White test takers (M = .67, SD = .56) when the nonfocal actor was not Black (i.e., White or Asian), t(186) = -4.00, p < .001. The effect size of this
difference is large by Cohen’s standards.

Table 26. Helping item 2 score differences between Black and White test takers
Condition
Test Taker Race
M
SD
t
.77
.73
Black test taker (n = 13)
Black non-focal actor
0.33
.71
.57
White test taker (n = 83)
.11
.58
Black test taker (n = 18)
Asian or White non-focal actor
-4.00**
.56
White test taker (n = 170) .67
Note. **p < 0.001.

d
--1.00

Next, the findings in this study are discussed along with some of their practical
implications. Study limitations and future directions for research are likewise considered. Before

135

proceeding, however, the results of the exploratory analyses in this section are summarized in
Table 27 below.

136

Table 27. Additional analyses of effects of demographic cues on performance

Test
Is there a four way interaction of
respondent race, respondent gender,
race match between respondent and
non-focal actor, and gender match
between respondent and non-focal
actor?
Which item(s) explain the two-way
interaction of respondent race and
race match on total SJT score (i.e.,
which item(s) do Black test takers
perform differently depending on race
match)?
Do Black test takers perform worse
than White test takers when the nonfocal actor is Black?
Do Black test takers perform worse
than White test takers when the nonfocal actor is not Black?

Finding
DV: Helping DV: Helping
situation 1
situation 2
score
score

DV: SJT score

DV: Conflict
situation 1
score

DV: Conflict
situation 2
score

No; two-way interaction of
respondent race and race
match between respondent
and non-focal actor

--

--

--

--

--

No

Yes

No

No

--

--

No

--

--

--

--

Yes

--

--

137

DISCUSSION
The purpose of this study was to consider the implications that demographic cues in
video-based situational judgment tests may have for how test takers of different demographic
profiles (gender, racial) respond to the SJT items as well as how they react to the test. Two types
of items, conflict and helping, were investigated. Hypotheses based on research on relational
demography, conflict, helping, and applicant reactions were tested.
Study Findings
Helping. Almost no support was found for hypotheses pertaining to how test takers
would respond to helping situations depending on their own as well as the non-focal actors’ race
and gender. The expectation that the effects of race mismatch on helping would be more negative
for White respondents viewing a minority non-focal actor than for minority respondents viewing
a White non-focal actor, based on relational demography research showing asymmetrical effects
of demographic similarity for high and low status groups, was not met. There was also no
support for hypotheses, likewise based on asymmetrical effects found in relational demography
research, about particular respondent groups offering specific non-focal actors lower levels of
help relative to levels of help under other combinations of respondents with targets. Also,
Asians, the ―model minority‖ (Alvarez et al., 2006), who are stereotypically believed to have
high work ethic, were not found to receive the highest level of help, in contrast to expectations.
There was partial support for the hypothesis that female respondents will offer more help to nonfocal actors than will male respondents, and this finding fits in with research showing that
women, as compared to men, tend to perform more ―altruistic‖ behaviors targeted toward
individuals needing assistance (Eagly, 2009).
Conflict. Hypotheses pertaining to conflict received almost no support. It was not the
case that respondents endorsed more assertive and cooperative responses when seeing a non138

focal actor of the same race. This expectation had been based on findings that groups will
endorse a conflict avoidance strategy less so when considering conflict with those of the same
race (Davidson, 2001). Likewise contrary to expectations, respondents’ race and race match with
the non-focal actor did not interact to predict respondents’ preference for third party intervention
in conflict situations. Surprisingly, there were no differences in cooperativeness and
assertiveness shown by male and female respondents, despite studies on gender norm
expectations that would suggest differences should be seen (e.g., Graves & Elsass, 2005; Wilson
et al., 2003). On one response option, both men and women tended to be less assertive and more
cooperative in response to female non-focal actors. This support for my hypothesis, though
minimal, fits in with findings of benevolent sexism (Wilson & Gallois, 1985) and more positive
behavior (Carli, 1989) toward women. There was no support for hypotheses about certain groups
responding to specific non-focal actors with higher or lower levels of assertiveness,
cooperativeness, or preference for intervention relative to observed levels given other
combinations of respondents with targets. Finally, respondents’ gender and gender match with
the non-focal actor interacted to predict preference for third party intervention in conflict
situations, but only in the case of one response option (both male and female respondents were
less willing to get a third party to intervene in an unglued cartons situation when the non-focal
actor was female).
Examining reasons for lack of support for hypotheses. Although only a small set of
the helping and conflict hypotheses were supported (and in most cases just partially so), this is
due in part to the inability to test most of these at the level of dimensions of responses as had
been planned. One of the hypotheses that could be tested at the dimension level, that male
respondents would be more willing to help female non-focal actors than male ones while female

139

respondents would not distinguish between male and female actors when offering help, stands
out as having been supported. This finding is very much in line with meta-analytic findings on
the role of gender in helping (Eagly & Crowley, 1986).
It should be acknowledged, however, that there remains the possibility that most of the
hypotheses would still not have been supported even if they could have been tested as planned.
Ideas from relational demography may have limited application to the context of test taker
responses on a video-based SJT. Relational demography takes into consideration the different
profiles of those who interact with one another to make predictions of outcomes for the
interacting parties. The theory was not completely applied in the case of this research in the
sense that the focal actor in each video was left out of the equation. It is possible that respondents
took not only their own and the non-focal actor’s demographics into consideration when
responding to the test question, but also incorporated information on the demographics of the
focal actor (White female) into their decisions of how to respond. It cannot be ruled out, for
example, that some respondents actually assumed the role of the focal actor in the situation,
taking on her demographic profile, to respond to a given situation, or that some respondents,
without actually assuming the focal individual’s role, simply provided what they believed to be a
desirable response from the standpoint of a White female. This research may have produced
more significant findings if it had been possible to create situations that featured just one actor,
while only implying the presence of the other one (see MacDorman, Coram, Ho, & Patel, 2010
for an example). In that case, there would be no physically present ―focal‖ actor to take into
consideration. The relational demography perspective could then be more precisely mapped onto
the ―interaction‖ of the respondent and actor.

140

Applicant reactions. Surprisingly, the findings pertaining to applicant reactions were
mostly inconsistent with research suggesting that minorities (e.g., Blacks, women) react
positively to diversity (e.g., Avery et al., 2004). The only exception to this was that minorities
(Asians, Blacks) did perceive better opportunity to perform when they viewed minority nonfocal actors. This limited support of my reactions hypotheses is further discussed in the
limitations section.
Response effectiveness. Arguably, the most interesting findings in this study were the
more exploratory ones pertaining to response effectiveness (i.e., actual performance on the test).
As a group, minority and White respondents performed somewhat worse on helping item 1 when
the non-focal actor did not match them in race. Only minority respondents performed worse on
helping item 2 when the non-focal actor did not match them in race. It was because of this
particular item that Black respondents scored higher on the SJT when the non-focal actor was
Black as opposed to White or Asian. There was also a significant three-way interaction of
respondent sex, non-focal actor sex, and non-focal actor race on helping item 1 score. With a
White non-focal actor (but not a Black one), men and women scored differently depending on
the actor’s gender. Male and female respondents got equal scores on helping item 1 when the
White non-focal actor was female. But when the White non-focal actor was male, female
respondents scored higher on the item than male respondents. This three way interaction was not
observed on helping item 2 though. Finally, minority respondents performed better on both
conflict items when the non-focal actor was female, while White respondents were not
influenced by non-focal actor gender in these items.
The theoretical background for this study suggests that differences in item responses may
occur based on demographic cues in the test. However, a substantive analysis of the link between

141

item scores and most and least likely responses to these items (i.e., the basis for determining item
scores) could not be done in most cases. This makes it difficult to interpret substantively the
effects that were found. Furthermore, some of the inconsistency in effects found across the same
type of items (i.e., helping) makes it even harder to offer a theoretical explanation. It should be
noted, as well, that because the items were not written for the constructs of interest, any theory
would have to be imposed on the items. Larger sample sizes would have facilitated substantive
analyses of score differences. However, the limited ability to offer theoretical interpretations for
some of the findings should not eliminate their practical value.
Practical Implications
The finding that Black test takers scored lower on average than White test takers on
helping item 2 when the non-focal actor was Asian or White, but not when the non-focal actor
was Black, has special practical relevance. It suggests that test developers potentially have some
amount of control over test bias when designing video-based tests. Yet, this issue needs to be
further investigated to determine the extent to which the current finding is generalizable to other
types of items.
Given the findings in this study it is appropriate to ask how large an issue for
consideration demographic cues in video based tests poses for test developers. The appropriate
answer is arguably ―it depends.‖ Current findings do show demographic cues at times having an
influence both on responses and actual performance on the test, as well as on minority
respondents’ perceived opportunity to perform. The extent to which companies may be
compelled to take these initial findings into consideration in developing video-based test
materials may depend on factors such as the diversity of their applicant pool (e.g., does the
company have a lot of difficulty hiring enough minorities?) and importance placed on projecting

142

a positive image of inclusion to minority applicants (i.e., does the company make it a goal to
convey organizational diversity?). The competencies assessed by a video-based test would also
be important to consider. The current findings speak to demographic cue effects that may occur
for interpersonal types of situations having to do with conflict and helping between coworkers.
Any results are therefore of more relevance to companies assessing these types of competencies.
Limitations and Future Directions
Although this study makes significant inroads into an area of study that past research has
offered little insight into, and provides some interesting initial findings with regard to the impact
of demographic cues in video-based situational judgment tests on responses to, performance on,
and reactions to these types of tests, it does have a number of limitations that should be
discussed. Some of these have to do with sampling, inherent multidimensionality of situational
judgment tests, context and administration factors, the set of variables that were chosen for
inclusion in (versus exclusion from) this research, and limited sampling from the domain of SJT
item types.
Sample size. Although a lot of time and effort was invested into getting enough men and
minorities for this study, this effort was more successful with regard to the former than the latter
demographic. Numbers of Asian and Black respondents were relatively small in this study and
particularly in the case of the four-way interactions I tested, there were really too few to find
significant effects even if these effects were there to be found. Thus, I have little confidence in
the null findings with regard to any of the four-way interactions. This area of research would
yield well to studies that have much larger numbers of minorities to sample from.
Multidimensionality. I tried to discover theoretical dimensions in responses to the SJT
items. However, these items were not designed to measure the dimensions I tried to pull out. As

143

described earlier, item stems were based on critical incidents technicians provided and response
options were actions these technicians proposed as effective or ineffective ways of responding to
the incidents. Given this non-theoretical approach to test development, it is not surprising that the
dimensions I hoped to find were not reliably measured by the SJT response options. The
multidimensional nature of the SJT items in this study meant that I could not test my hypotheses
about dimensions of responses as planned. Even though several of the hypotheses about helping
as a dimension could be tested, this helping score was ultimately composed of just a fourth of the
response options I meant for it to represent. Thus, most hypotheses had to be tested at the level of
the response options instead of dimensions and my ability to provide a compelling explanation
for why respondents may have endorsed a particular option more or less strongly depending on
their own race/gender, or an interaction of their own demographics with demographic cues in the
video, was limited. It should be pointed out, however, that the problem of failing to derive
construct validity for the SJT items in my study is not unique, but typical to SJT research. Most
SJTs are developed via non-theoretical approaches and little attention is paid to constructs
assessed by these tests (Christian, Edwards, & Bradley, 2010 provide a review). Though it is
likely to be difficult, future research can investigate the effects of demographic cues on SJT
items developed via a theoretical approach so as to have the ability to draw stronger theoretical
conclusions from any findings.
Weights. Response item weighting in this study may be argued to be a limitation. Even
though I was able to establish adequate agreement among judges to use the means of ratings they
provided as weights for likelihood ratings respondents assigned to response options, the adequate
agreement came with the collapsing of the five-point scale judges used, into a three-point scale.
Even utility of the practice of weighting by means can be questioned as many will argue that unit

144

weights would have been adequate (see Bobko, Roth, & Buster, 2007 for a discussion). As I also
tried to apply unit weights as an alternative to weighting by means and the results of analyses
ended up about the same, the approach did not make a difference in this study. Ultimately,
weighting was done for the sake of one dimension of one SJT item so the issue of weights
actually has limited bearing on the findings in this study.
Context and administration. This study examines in a lab setting a type of test that in
practice would be administered in a selection context. While the experimental nature of this
study is a strength in that variables of interest could be isolated in the lab, it is simultaneously a
weakness in that high stakes testing is associated with generally higher test taker motivation and
effort than typically seen in a lab setting. Further, SJTs, including the one items for this study
were drawn from, are typically administered as timed tests. For the original test respondents had
70 seconds to answer an item (and there was a ten second notification built in). As the 4-item test
in my study was not timed, respondents had as much time as they wanted to deliberate before
providing their responses. It is not possible to know how the results of this study may have been
different given a timed selection context where test takers would be more motivated and forced
to make faster decisions. A direction for future research may be to consider the moderating
effects that motivational and administration-related factors (e.g., time, response instructions)
could have on the relationship of demographic cues in SJTs with test takers’ responses.
The lab context may have also contributed to the limited findings with regard to applicant
reactions. Respondents might have perceived the instructions to ―think of yourself as a job seeker
applying for a technician job with ABC Corporation,‖ as too artificial. They may also not have
recognized the connection between the job description for a plant technician and the
interpersonal content of the test they were taking ―to see if you will be further considered by

145

ABC Corporation for a technician position,‖ although an attempt was made to make that
connection clear in the instructions (which respondents may or may not have fully read). If the
respondents perceived the reactions questions to lack relevance, this overarching impression may
have driven their responses to these questions more so than demographic cues in the test. On
average, test takers responded somewhere between ―somewhat disagree‖ and ―neither disagree
nor agree‖ to scales for opportunity to perform, and job relatedness (both predictive and content).
They averaged at about ―neither disagree nor agree‖ on the compatibility scale and between
―neither disagree nor agree‖ and ―somewhat agree‖ on the attraction scale. These results do seem
to suggest that respondents may have perceived the questions to be of limited relevance, and
therefore did not have particularly strongly reactions to them. Future research may add to current
findings by making an effort to maximize ecological validity when investigating applicant
reactions to demographic cues in tests.
Categorization process. In this study, hypotheses predicting effects of demographic cues
on test takers’ response tendencies assumed that a categorization process occurs for respondents
when they view actors of a certain race and gender. Based on the finding that respondents
sometimes were not able to correctly identify the demographic profile(s) of the actor(s) they
viewed, there is reason to question whether all respondents were able to categorize actors into
their ―correct‖ demographic groups. Aside from reasons of inattention or purposeful error, there
are two broad possibilities as to why these categorization errors may have occurred. The first
possibility is that some respondents, though able to correctly categorize the actors mentally, were
later unable to assign correct labels to these actors when asked (e.g., some may have been
confused about what the label Middle Eastern actually means) or were unwilling to do so (e.g.,
recent media reports have discussed contemporary youths’ tendency toward avoiding placing

146

themselves or others into narrow demographic categories; Saulny, 2011a; 2011b). The second
possibility is that some respondents were unable to even categorize the actors mentally. This
latter scenario presents a bigger problem as it has the categorization process actually not
occurring. It is not possible to know the extent to which the test takers in this study experienced
one or the other problem. Yet, as respondents who failed the manipulation check(s) were
excluded from all analyses, as were multiracial and ―other‖ respondents from analyses pertaining
to race, the influence of this issue of categorization failure on the results of this study should
have been minimized. An interesting direction for future research may be to consider the
implications that changing trends in demographic categorization of self and others has for how
test takers perceive and respond to others in the context of selection testing (both in terms of test
responses and applicant reactions).
Also related to the issue of categorization, it should be noted that Asian respondents in
this study were a much less homogenous group than the White and Black respondent groups. For
example, both East Asians and Indians were part of the ―Asian‖ group of respondents in this
study. While Indian respondents most likely did not see the East Asian non-focal actors in the
videos as belonging to the same demographic group as them, this distinction could not be made
in the analyses pertaining to non-focal actor race and race match between respondent and nonfocal actor, as respondents were not asked to report what region of the Asian continent they
associate themselves with. Those analyses may be muddied then and perhaps true effects of
racial cues for East Asian respondents were masked because of this. It would be important for
future research in this area to address this limitation by better identifying when respondents are
actually paired with non-focal actors of the same demographic profile as them. Such research

147

would be able to shed better light on whether Asian respondents are sensitive to demographic
cues in video-based tests.
Limited demographic profiles of actors. To keep the design of this study manageable,
decisions had to be made regarding which demographic cues to focus on. The decision was made
to keep the age of the actors constant and to limit non-focal actor race to three groups only. As a
result, this study cannot offer findings regarding how age may interact with other demographic
cues to influence responses to the SJT and how respondents may react to Hispanic actors in
situational judgment items, for example. This limits the practical implications of this study as
organizations are likely to develop SJTs that show a wider range of ages and races (so as to
convey diversity) than I examined. Other research can examine additional demographic cues as
predictors of test takers’ responses to SJTs. As this study also kept the race, gender, and age of
the focal actors constant, investigating possible effects of focal actor (i.e., again, the one the test
taker needs to assume the role of) demographics on test taker outcomes is likewise a topic for
future research to consider.
Limited demographic profiles of respondents. Respondents in this study were less
diverse in terms of their demographics (e.g., age, race, education, job experience) than is the
typical organization’s applicant pool. The findings of this study do not address, for example,
whether more experienced test takers may respond similarly to demographic cues in SJTs with
regard to response tendencies and actual performance. It is possible, for instance, that those with
relevant job experience will be more likely to respond effectively to SJT questions regardless of
the demographic cues that may be present in the test. Future research can look into these types of
questions.

148

Item type and test length. The generalizability of the current findings (and in certain
cases lack thereof) may be limited by the fact that, due to resource limitations, only 4 items were
included in this study. Further, they were just two types of interpersonal items (i.e., conflict,
helping), albeit very common ones, out of a wider range of possible SJT item content types (e.g.,
handling a technical work problem, delivering bad news to a subordinate, providing
encouragement to a coworker). The findings, particularly the practical ones with regard to
performance differences of Black test takers depending on race cues in the test, may or may not
generalize to other types of items and longer tests. This question in particular would be important
to investigate further given the important dilemma of adverse impact in high stakes testing and
its implications for diversity in hiring.
Overview. Despite the limitations in this research study, it makes an important
contribution to the literature by applying a relational demography perspective to examining
responses to a video-based SJT. This study adds to researchers’ knowledge of the relevance
relational demography has for issues in selection (e.g., recruitment; Avery, Hernandez, & Hebl,
2004; Goldberg, 2005; Perkins, Thomas, &Taylor, 2000; interviews; Buckley et al., 2007). It fits
in well with the extensive research that examines sources of construct-irrelevant variance in SJT
performance (e.g., the extent to which characters look/act human; MacDorman et al., 2010;
administration medium; Lievens & Sackett, 2006; faking; Peeters & Lievens, 2005; response
instructions; Nguyen & McDaniel, 2003; Ployhart, & Ehrhart, 2003) as well as other types of
tests (e.g., cognitive ability; Brown & Day, 2006; Schmitt, 1988; personality; Ellingson, Sackett,
& Hough, 1999; math/mechanical ability; Grand, Ryan, Schmitt, & Hmurovic, 2011; Hofstetter,
2003). This cited line of research typically considers the extent to which, given the presence of
construct irrelevant variance, tests still predict outcomes of interest. Examining this issue was

149

beyond the scope of the current research. This would, however, be a useful direction for the
current research to go in. Thus, given that demographic cues were found to impact test
performance in some cases, it would be important to examine how the criterion-related validity
of video-based SJTs may vary depending on the interaction of test characteristics (actor race,
gender) with test taker characteristics (respondent race, gender). As discussed earlier, as research
moves forward, a broader set of test and test taker characteristics might be useful to consider
than was included in the current study.
Future research should continue to apply a relational demography perspective to videobased, and other types of SJTs that use a visual medium, where test takers have the opportunity
to view the actors or characters involved. The newest trend in SJT development is the use of
avatars instead of human actors. Some of the same relationships examined in the current study
would be relevant to examine using avatars as they, like human actors, have demographic
profiles. The use of avatars may be a particularly promising direction for SJTs with regard to test
fairness in that avatar SJTs are more amenable than human actor SJTs to customization
according to the needs or preferences of the test taker (i.e., there is no need to recreate the test if
changing a character’s race/gender/age is desired). Research will need to examine this new
medium further to determine how test taker performance and reactions may be optimized via
customization. This direction of research would have a lot of practical value for test developers
and organizations.
Conclusion
This study applied a relational demography perspective to examine differential effects of
demographic cues in a video-based situational judgment test on test takers belonging to different
race and gender groups. Findings suggest some utility for looking at test performance and

150

reactions from the standpoint of the interaction of respondents’ and actors’ demographic profiles.
Further research looking into ways in which actor-related elements of video-based test design
may be used to enhance particular test taker groups’ performance and reactions seems warranted.

151

APPENDICES

152

APPENDIX A
Text Preceding SJT Test (Adapted from Bauer et al., 2001)
Please think of yourself as a job seeker applying for a technician job with ABC Corporation.
This company is offering a yearly salary 20% higher than other companies you’re considering
and is located in town you like. In talking with people hired in the last 5 years, you have
discovered that employees have received an average of three promotions in that time. The
company also has been rated as a leader in the industry in terms of proactive environmental
policies and was rated as one of the top 100 places to work by US News & World Reports.
Today you will be taking an initial screening test to see if you will be further considered by ABC
Corporation for a technician position within their company. It is a test of interpersonal ability.
You will be viewing and responding to a number of scenarios reflecting interpersonal situations
that may be encountered on the technician job. For each scenario you view, you will be asked to
provide your most as well as least likely course of action in that situation.
After the test you will be filling out several short questionnaires. Thank you for your
participation.

153

APPENDIX B
Job Description for an Entry Level Plant Technician
Plant technicians are hired at entry level as members of a work team. Technician work teams are
responsible for keeping production flowing, for meeting performance goals and for maintaining
the high level of quality that consumers expect from the company’s products.
Technicians do various types of work, including running high-speed converting and packing
equipment, controlling the process from a computer terminal, and performing preventive
maintenance on equipment. Technicians are expected to get involved, take charge of situations
and confront business-related problems.
Technicians generally have strong technical, interpersonal and problem-solving skills.
The typical plant operates 7-days per week. Technicians work any day of the week, including
weekends. The hours of work vary from 8 to 12 hours per day. Shift rotations vary from
working a single straight shift to working two or three rotating shifts. Technicians must be
willing to rotate and to work any shift.

154

APPENDIX C
Situational Judgment Test Items

Note:
1. The “Other Person” was one of three races (White, Black, or Asian) and either male or
female depending on the study condition. Response options referred to the Other Person as “he”
and “him” when that actor was male and as “she” and “her” when that actor was female. The
term “Other Person” was not used during the experiment.
2. “Focal Person” was always a White female. The Focal Person in each scenario had a name
(the names were Alice, Jane, Carol, and Ann); the term “Focal Person” was not used during the
experiment.

Conflict Situation 1 (Notes: Competence stereotype is relevant; goal of intervention would be to
report coworker’s poor performance)
Focal Person: What’s going on? This is the fourth one in a row that hasn’t been glued shut. Oh,
I better call over.

Focal Person (on phone): Yeah, this is Alice in shipping. Yeah, you guys are sending over
cartons that aren’t glued shut. So, you better shut down that line and figure out what your
problem is.
Other Person: Hey, I’m standing here looking at the line and every case is perfect. Our case
packer is working just fine.

Please rate each response option according to how likely you would be to choose that response
if you were Focal Person.
If you were Focal Person, what would you MOST and LEAST likely do?
a. Tell your coworker that you will continue to watch for unglued cartons and get back
to him (her).
b. Assume that the cartons are being glued properly because your coworker hasn’t
noticed the problem.
c. Insist that there must be a problem your coworker hasn’t noticed since unglued
cartons are arriving at the warehouse.

155

d. Ask another team member to discuss the problem with your coworker.
e. Let your team leader know that your coworker is allowing unglued cartons to get to
the warehouse.

Help Seeking Situation 1 (Notes: Competence stereotype is relevant)
Other Person: Carol, can I bug you for a minute?

Focal Person: Yeah, what?
Other Person: I’m having a real problem with Frank. I don’t think he listens to me. He doesn’t
think I know anything. I was hoping you might give me some ideas on how to deal with him
better?

Please rate each response option according to how likely you would be to choose that response
if you were Focal Person.
If you were Focal Person, what would you MOST and LEAST likely do?

a. Explain to your coworker that you are very busy, but if he (she) wants to talk later,
you’ll tell him (her) how you handle Frank.
b. Suggest that your coworker ignore Frank because working with Frank can be
difficult.
c. Tell your coworker that he (she) needs to learn how to handle his (her) own conflicts
with Frank.
d. Tell your coworker how you deal with Frank.
e. Suggest that your coworker meet with Frank to talk about their differences.
f. Suggest that your coworker talk to the team leader about his (her) problem with
Frank.

Conflict Situation 2 (Notes: Stereotypes pertaining to aggressiveness and agreeableness are
relevant; goal of intervention would be reconciliation)
Focal Person: Hey, can I sit here for a minute.
156

Other Person: Sure. Sit there all day if you like. I was just leaving.
Focal Person: Wait a minute. I want to find out why you’re always giving me a hard time.
Other Person: Look, I only give people a hard time if they make mistakes. Just don’t make
mistakes and we’ll get along fine.

Please rate each response option according to how likely you would be to choose that response
if you were Focal Person.
If you were Focal Person, what would you MOST and LEAST likely do?

a. Ask your coworker what you could do to get along better.
b. Ask the team leader for suggestions about how to work with your coworker.
c. Try to avoid your coworker whenever possible because you can’t change him (her).
d. Tell your coworker that he (she) should act like a team member and try to work
together.
e. Ask the team leader to meet with you and your coworker to work out your
differences.
f. Ask other team members for suggestions on how to work with your coworker.

Help Seeking Situation 2 (Notes: Work ethic stereotype is relevant)
Other Person: Hey Ann, can you do me a real big favor? I want to take a training class next
week but I’m on night shift. Can you switch with me?

Focal Person: I took nights for you like a month ago so you could take some training class,
didn’t I?

Other Person: Yeah, but this is one the new I-3000 wrapping equipment and I think it would be
real helpful. Come on, can you switch with me again?

157

Focal Person: Hey, I don’t even think you’ve used the training you got the last time. And quite
frankly, I don’t like working nights anymore than you do. And it seems to me like your always
coming with some new idea on how not to work nights, man.
Other Person: Listen, we’re supposed to help out members of our team, right? I already got
permission to sign-up and you’ve already had this training. I think you should give me the
chance to learn something that’s going to help us do a better job.

Please rate each response option according to how likely you would be to choose that response
if you were Focal Person.
If you were Focal Person, what would you MOST and LEAST likely do?
a. Agree to switch with your coworker so that he (she) can take the training course.
b. Agree to switch with your coworker, but only if he (she) will switch with you in a few
weeks.
c. Tell your coworker you think he (she) is taking advantage of the system and you
won’t switch.
d. Tell your coworker you won’t switch, and he (she) should wait until training is
offered at another time that doesn’t interfere with his (her) schedule.
e. Tell your coworker that you think he (she) is taking advantage of the system, but you
will help him (her) find someone else to switch.

158

APPENDIX D
Reactions Items Pertaining to Opportunity to Perform (Adapted from Bauer et al., 2001)
1. I could really show my skills and abilities through this test.
2. This test allowed me to show what my interpersonal skills are.
3. This test gives individuals the opportunity to show what they can really do.
4. I was able to show what I can do on this test.

159

APPENDIX E
Reactions Items Pertaining to Job-Relatedness (Adapted from Bauer et al., 2001)
Job Relatedness—Predictive
1. Doing well on this test means a person can do the technician job well.
2. A person who scored well on this test will be a good technician.
Job Relatedness—Content
1. It would be clear to anyone that this test is related to the technician job.
2. The content of the test was clearly related to the technician job.

160

APPENDIX F
Reactions Items Pertaining to Attraction (Adapted from Perkins et al., 2000)
1. I would request additional information regarding the possibility of employment with this
company.
2. I would like this company to recruit on campus.
3. I would speak to a company representative about the possibility of employment.
4. I would think this organization is attractive.
5. I would not recommend this company to a friend. (Reverse coded)
6. I would like this organization.

161

APPENDIX G
Reactions Items Pertaining to Compatibility (Adapted from Perkins et al., 2000)
1. I would feel at home working for an organization like this.
2. I would very much like to work for this organization.
3. This organization will likely meet my desires and needs.
4. I would have no problems adjusting to this organization.

162

APPENDIX H
Demographic Questions
1. What is your age? _______
2. What is your gender?
a. Male
b. Female
3. Is your ethnicity Hispanic/Latino?
a. Yes
b. No
4. If your ethnicity is not Hispanic/Latino please select your race below. Select more than 1
response if applicable.
a. American Indian or Alaska Native
b. Asian
c. Black or African American
d. Native Hawaiian or Other Pacific Islander
e. White
f. Other
5. If other, please specify. ___________
6. What is your class standing?
a. Freshman
b. Sophomore
c. Junior
d. Senior
e. Other/non-degree
7. What is your cumulative college GPA?
a. less than 2.00
b. 2.00 to 2.29
c. 2.30 to 2.59
d. 2.60 to 2.89
e. 2.90 to 3.19
f. 3.20 to 3.49
g. 3.50 to 3.79
h. 3.80 to 3.99
i. 4.00
8. Which of the following best describes your intended or current major?
a. None, I have not yet decided on a major
b. Business
c. Engineering
163

d. Fine Arts or Humanities
e. Social Science (i.e., Psychology, Sociology, Political Science, etc)
f. Natural Science (i.e., Biology, Chemistry, Physics, etc)
g. Other

164

APPENDIX I
Distributions of Responses for Those Who Failed Manipulation Checks

Table 28. Distribution of responses for those who failed the manipulation check for conflict situation 1
East
East
Middle
Middle
Asian
Asian
Black
Black Caucasian Caucasian Eastern Eastern
Condition
Female
Male Female Male
Female
Male
Female
Male
X
-----100%
-Asian Female non-focal
-X
-----50%
Asian Male non-focal
--X
50%
----Black Female non-focal
-7.7%
-X
---76.9%
Black Male non-focal
----X
---White Female non-focal
----75%
X
--White Male non-focal
Note. X indicates the correct response to the manipulation check for conflict situation 1 given a particular condition.

Table 29. Distribution of responses for those who failed the manipulation check for conflict situation 2
East
East
Middle Middle
Asian
Asian
Black
Black Caucasian Caucasian Eastern Eastern
Condition
Female
Male Female Male
Female
Male
Female
Male
Asian Female non-focal
X
-----50%
-Asian Male non-focal
-X
-----66.7%
Black Female non-focal
--X
-----Black Male non-focal
---X
----White Female non-focal
12.5%
---X
---White Male non-focal
-9.1%
---X
--Note. X indicates the correct response to the manipulation check for conflict situation 1 given a particular condition.

165

Cannot
tell
-50%
50%
15.4%
100%
25%

Total
N
4
2
2
13
4
4

Cannot
tell
50%
33.3%
--87.5%
90.9%

Total
N
6
3
0
0
8
11

Table 30. Distribution of responses for those who failed the manipulation check for helping situation 1
East
East
Middle Middle
Asian
Asian
Black
Black Caucasian Caucasian Eastern Eastern
Condition
Female
Male Female Male
Female
Male
Female
Male
Asian Female non-focal
X
100%
------Asian Male non-focal
50%
X
-----25%
Black Female non-focal
--X
-----Black Male non-focal
---X
----White Female non-focal
20%
---X
---White Male non-focal
-----X
--Note. X indicates the correct response to the manipulation check for helping situation 1 given a particular condition.

Table 31. Distribution of responses for those who failed the manipulation check for helping situation 2
East
East
Middle Middle
Asian
Asian
Black
Black Caucasian Caucasian Eastern Eastern
Condition
Female
Male Female Male
Female
Male
Female
Male
Asian Female non-focal
X
-----66.7%
-Asian Male non-focal
-X
-----66.7%
Black Female non-focal
--X
---100%
-Black Male non-focal
--11.1%
X
---33.3%
White Female non-focal
----X
14.3%
--White Male non-focal
----75%
X
--Note. X indicates the correct response to the manipulation check for helping situation 2 given a particular condition.

166

Cannot
tell
-25%
-100%
80%
100%

Total
N
1
4
0
1
5
1

Cannot
tell
33.3%
33.3%
-55.6%
85.7%
25%

Total
N
3
3
1
9
7
4

REFERENCES

167

REFERENCES

Alvarez, A. N., Juang, L., & Liang, C. T. H. (2006). Asian Americans and racism: When bad
things happen to ―model minorities.‖ Cultural Diversity and Ethnic Minority Psychology,
12, 477-492.
Auster, C. J., & Ohm, S.C. (2000). Masculinity and femininity in contemporary American
society: A reevaluation using the Bem Sex-Role Inventory. Sex Roles, 43, 499-528.
Avery, D. R. (2003). Reactions to diversity in recruitment advertising—Are differences Black
and White? Journal of Applied Psychology, 88, 672-679.
Avery, D. R., Hernandez, M., & Hebl, M. R. (2004). Who’s watching the race? Racial salience
in recruitment advertising. Journal of Applied Social Psychology, 34, 146-161.
Avery, D. R., & McKay, P. F. (2006). Target practice: An organizational impression
management approach to attracting minority and female job applicants. Personnel
Psychology, 59, 157-187.
Banki, S., & Latham, G. P. (2010). The criterion-related validities and perceived fairness of the
situational interview and the situational judgment test in an Iranian organisation. Applied
Psychology: An International Review, 59, 124-142.
Bauer, T. N., Truxillo, D. M., Paronto, M. E., Weekley, J. A., & Campion, M. A. (2004).
Applicant reactions to different selection technology: Face-to-face, interactive voice
response, and computer-assisted telephone screening interviews. International Journal of
Selection and Assessment, 12, 135-148.
Bauer, T. N., Truxillo, D. M., Sanchez, R. J., Craig, J. M., Ferrara, P., & Campion, M. A. (2001).
Applicant reactions to selection: Development of the selection procedural justice scale
(SPJS). Personnel Psychology,54, 387-419.
Bergman, M. E., Drasgow, F., Donovan, M. A., Henning, J. B., & Juraska, S. E. (2006). Scoring
situational judgment tests: Once you get the data, your troubles begin. International
Journal of Selection and Assessment, 14, 223-235.
Berscheid, E., & Walster, E. (1974). Physical attractiveness. Advances in Experimental Social
Psychology, 7, 157-215.
Bobko, P., Roth, P. L., & Buster, M. A. (2007). The usefulness of unit weights in creating
composite scores: A literature review, application to content validity, and meta-analysis.
Organizational Research Methods,10, 689-709.
Brewer, M. B., & Harasty Feinstein, A. S. (1999). Dual processes in the cognitive representation
of persons and social categories. In S. Chaiken, & Y. Trope (Eds.), Dual process theories
in social psychology (pp. 255–270). New York: Guilford Press.
168

Brooks, M. E., & Highhouse, S. (2006). Can good judgment be measured? In J. A. Weekley &
R. E. Ployhart (Eds.), Situational judgment tests (pp. 39-55). San Francisco: Jossey-Bass.
Brown, R. P., & Day, E. A. (2006). The difference isn’t Black and White: Stereotype threat and
the race gap on Raven’s Advanced Progressive Matrices. Journal of Applied Psychology,
91, 979-985.
Browne, I., & Misra, J. (2003). The intersection of gender and race in the labor market. Annual
Review of Sociology, 29, 487-513.
Brumbaugh, A. M. (2009). Why do I identify with thee? Let me count three ways: How ad
context influences race-based character identification. Psychology & Marketing, 26, 970986.
Buckley, M. R., Jackson, K. A., Bolino, M. C., Veres, J. G. III., & Feild, H. S. (2007). The
influence of relational demography on panel interview ratings: A field experiment.
Personnel Psychology, 60, 627-646.
Carli, L. L. (1989). Gender differences in interaction style and influence. Journal of Personality
and Social Psychology, 56, 565-576.
Chan, D., & Schmitt, N. (1997). Video-based versus paper-and-pencil method of assessment in
situational judgment tests: Subgroup differences in test performance and face validity
perceptions. Journal of Applied Psychology, 82, 143-159.
Chan, D., & Schmitt, N. (2002). Situational judgment and job performance. Human
Performance, 15, 233-254.
Chatman, J. A., & Flynn, F. J. (2001). The influence of demographic heterogeneity on the
emergence and consequences of cooperative norms in work teams. The Academy of
Management Journal, 44, 956-974.
Chatman, J. A., & O’Reilly, C. A. (2004). Asymmetric reactions to work group sex diversity
among men and women. Academy of Management Journal, 47, 193-208.
Chattopadhyay, P., Tluchowska, M., & George, E. (2004). Identifying the ingroup: A closer look
at the influence of demographic dissimilarity on employee social identity. Academy of
Management Review, 29, 180-202.
Christian, M. S., Edwards, B. D., & Bradley, J. C. (2010). Situational judgment tests: Constructs
assessed and a meta-analysis of their criterion-related validities. Personnel Psychology,
63, 83-117.
Dalessio, A. T. (1994). Predicting insurance agent turnover using a video-based situational
judgment test. Journal of Business and Psychology, 9, 23-32.
Davidson, M. N. (2001). Know thine adversary: The impact of race on styles of dealing with
conflict. Sex Roles, 45, 259-276.
169

De Dreu, C. K. W., Evers, A., Beersma, B., Kluwer, E. S., & Nauta, A. (2001). A theory-based
measure of conflict management strategies in the workplace. Journal of Organizational
Behavior, 22, 645-668.
Deshpandé, R., & Stayman, D. M. (1994). A tale of two cities: Distinctiveness theory and
advertising effectiveness. Journal of Marketing Research, 31, 57-64.
Eagly, A. H. (2009). The his and hers of prosocial behavior: An examination of the social
psychology of gender. American Psychologist, 64, 644-658.
Eagly, A. H., & Crowley, M. (1986). Gender and helping behavior: A meta-analytic review of
the social psychological literature. Psychological Bulletin, 100, 283-308.
Eagly, A. H., Karau, S. J., & Makhijani, M. G. (1995). Gender and the effectiveness of leaders:
A meta-analysis. Psychological Bulletin, 117, 125-145.
Ellingson, J. E., Sackett, P. R., & Hough, L. M. (1999). Social desirability corrections in
personality measurement: Issues of applicant comparison and construct validity. Journal
of Applied Psychology, 84, 155-166.
Elsass, P. M., & Graves, L. M. (1997). Demographic diversity in decision-making groups: The
experiences of women and people of color. Academy of Management Review, 22, 946973.
Ensher, E. A., & Murphy, S. E. (1997). Effects of race, gender, perceived similarity, and contact
on mentor relationships. Journal of Vocational Behavior, 50, 460-481.
Fiske, S. T., Lin, M., & Neuberg, S. L. (1999). The continuum model, ten years later. In
S.Chaiken, & Y. Trope (Eds.), Dual process theories in social psychology (pp. 231–
254).New York: Guilford Press.
Foley, S., Linnehan, F., Greenhaus, J. H., & Weer, C. H. (2006). The impact of gender similarity,
racial similarity, and work culture on family-supportive supervision. Group &
Organization Management, 31, 420-441.
Gaines, S.O., Jr., Marelich, W.D., Bledsoe, K. L., Steers, W.N., Henderson, M.C., Granrose, C.
S., et al. (1997). Links between race/ethnicity and cultural values as mediated by
racial/ethnic identity and moderated by gender. Journal of Personality and Social
Psychology, 72, 1460–1476.
Giebels, E. & Yang, H. (2009). Preferences for third-party help in workplace conflict: A crosscultural comparison of Chinese and Dutch employees. Negotiation and Conflict
Management Research, 2, 344-362.
Glick, P., & Fiske, S. T. (2001). An ambivalent alliance: Hostile and benevolent sexism as
complimentary justifications for gender inequality. American Psychologist, 56, 109-118.

170

Goldberg, C., Riordan, C. M., & Zhang, L. (2008). Employees’ perceptions of their leaders: Is
being similar always better? Group & Organization Management, 33, 330-355.
Grand, J. A., Ryan, A. M., Schmitt, N., & Hmurovic, J. (2011). How far does stereotype threat
reach? The potential detriment of face validity in cognitive ability testing. Human
Performance, 24, 1-28.
Graves, L. M., & Elsass, P. M. (2005). Sex and sex dissimilarity effects in ongoing teams: Some
surprising findings. Human Relations, 58, 191-221.
Graves, L. M., & Powell, G. N. (1995). The effect of sex similarity on recruiters’ evaluations of
actual applicants: A test of the similarity–attraction paradigm. Personnel Psychology, 48,
85-98.
Graves, L. M., & Powell, G. N. (1996). Sex similarity, quality of the employment interview and
recruiters’ evaluation of actual applicants. Journal of Occupational and Organizational
Psychology, 69, 243-261.
Harrison, L. A., & Esqueda, C. W. (2001). Race stereotypes and perceptions about Black males
involved in interpersonal violence. Journal of African American Men, 5, 81-92.
Hausknecht, J. P., Day, D. V., & Thomas, S. C. (2004). Applicant reactions to selection
procedures: An updated model and meta-analysis. Personnel Psychology, 57, 639-683.
Herzog, S., & Oreg, S. (2008). Chivalry and the moderating effect of ambivalent sexism:
Individual differences in crime seriousness judgments. Law & Society Review, 42, 45-73.
Hinds, P. J., Carley, K. M., Krackhardt, D., & Wholey, D. (2000). Choosing work group
members: Balancing similarity, competence, and familiarity. Organizational Behavior
and Human Decision Processes, 81, 226-251.
Hodson, G., Dovidio, J. F., & Gaertner, S. L. (2002). Processes in racial discrimination:
Differential weighting of conflicting information. Personality and Social Psychology
Bulletin, 28, 460-471.
Hofstetter, C. H. (2003). Contextual and mathematics accommodation test effects for Englishlanguage learners. Applied Measurement in Education, 16, 159-188.
Hu, C., Thomas, K. M., & Lance, C. E. (2008). Intentions to initiate mentoring relationships:
Understanding the impact of race, proactivity, feelings of deprivation, and relationship
roles. Journal of Social Psychology, 148, 727-744.
Huddy, L., & Terkildsen, N. (1993). Gender stereotypes and the perception of male and female
candidates. American Journal of Political Science, 37, 119-147.
Huston, T. L., & Levinger, G. (1978). Interpersonal attraction and relationships. Annual Review
of Psychology, 29, 115-156.

171

Ibarra, H. (1992). Homophily and differential returns: Sex differences in network structure and
access in an advertising firm. Administrative Science Quarterly, 37, 422-447.
Jones, C., & Decotiis, T. A. (1986). Video-assisted selection of hospitality employees. The
Cornell H.R.A. Quarterly, 27, 68-73.
Judge, T. A., & Livingston, B. A. (2008). Is the gap more than gender? A longitudinal analysis
of gender, gender role orientation, and earnings. Journal of Applied Psychology, 93, 9941012.
Kidder, D., & McLean Parks, J. (2001). The good soldier: Who is s(he)? Journal of
Organizational Behavior, 22, 939-959.
King, E. B., Hebl, M. R., & Beal, D. J. (2009). Conflict and cooperation in diverse workgroups.
Journal of Social Issues, 65, 261-285.
Kirchmeyer, C. (1995). Demographic similarity to the work group: A longitudinal study of
managers at the early career stage. Journal of Organizational Behavior, 16, 67-83.
Koch, B. J., & Koch, P. T. (2007). Collectivism, individualism, and outgroup cooperation in a
segmented China. Asia Pacific Journal of Management, 24, 207-225.
Kozlowski, S. W. J., & Hattrup, K. (1992). A disagreement about within-group agreement:
Disentangling issues of consistency versus consensus. Journal of Applied Psychology, 77,
161-167.
Kraiger, K., & Ford, J. K. (1985). A meta-analysis of ratee race effects in performance ratings.
Journal of Applied Psychology, 70, 56-65.
Kravitz, D. A., Stone-Romero, E. F., and Ryer, J. A. (1997). Student evaluations of grade appeal
procedures: The importance of procedural justice. Research in Higher Education, 38,
699-726.
Kunda, Z., & Thagard, P. (1996). Forming impressions from stereotypes, traits, and behaviors: A
parallel-constraint-satisfaction theory. Psychological Review, 103, 284-308.
Kuntsman, J. W., & Plant, E. A. (2008). Racing to help: Racial bias in high emergency helping
situations. Journal of Personality and Social Psychology, 95, 1499-1510.
Landrine, H., Klonoff, E. A., Alcaraz, R., Scott, J., & Wilkins, P. (1995). Multiple variables in
discrimination. In B. Lott & D. Maluso (Eds.), The social psychology of interpersonal
discrimination (pp. 183-224). New York: Guilford Press.
Landy, F. J. (2008). Stereotypes, bias, and personnel decisions: Strange and stranger. Industrial
and Organizational Psychology: Perspectives on Science and Practice, 1, 379-392.
Latham, G. P., Saari, L. M., Pursell, E. D., & Campion, M. A. (1980). The situational interview.
Journal of Applied Psychology, 65, 422-427.
172

Lau, D. C., Lam, L. W., & Salamon, S. D. (2008). The impact of relational demographics on
perceived managerial trustworthiness: Similarity or norms? The Journal of Social
Psychology, 148, 187-208.
LeBreton, J. M., & Senter, J. L. (2008). Answers to 20 questions about interrater reliability and
interrater agreement. Organizational Research Methods, 11, 815-852.
Levin, S., Sinclair, S., Veniegas, R. C., & Taylor, P. L. (2002). Perceived discrimination in the
context of multiple group memberships. Psychological Science, 13, 557-560.
Lievens, F., Peeters, H., & Schollaert, E. (2008). Situational judgment tests: A review of recent
research. Personnel Review, 37, 426-441.
Lievens, F., & Sackett, P. R. (2006). Video-based versus written situational judgment tests: A
comparison in terms of predictive validity. Journal of Applied Psychology, 91, 11811188.
Liu, M. (2009). The intrapersonal and interpersonal effects of anger on negotiation strategies: A
cross-cultural investigation. Human Communication Research, 35, 148-169.
MacDorman, K. F., Coram, J. A., Ho, C-C., & Patel, H. (2010). Gender differences in the impact
of presentational factors in human character animation on decisions in ethical dilemmas.
Presence, 19, 1-17.
MacGeorge, E. L. (2003). Gender differences in attributions and emotions in helping contexts.
Sex Roles, 48, 175-182.
Martin, B. A. S., Lee, C. K-C., & Yang, F. (2004). The influence of ad model ethnicity and selfreferencing on attitudes: Evidence from New Zealand. Journal of Advertising, 33, 27-37.
McFarland, L. A., Ryan, A. M., Sacco, J. M., Kriska, S. O. (2004). Examination of structured
interview ratings across time: The effects of applicant race, rater race, and panel
composition. Journal of Management. 30, 435-452.
McDaniel, M. A., Morgeson, F. P., Finnegan, E. B., Campion, M. A., & Braverman, E. P.
(2001). Use of situational judgment tests to predict job performance: A clarification of
the literature. Journal of Applied Psychology, 86, 730-740.
McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in social
networks. Annual Review of Sociology, 27, 415-444.
Milliken, F. J., & Martins, L. L. (1996). Searching for common threads: Understanding the
multiple effects of diversity in organizational groups. Academy of Management Review,
21, 402-433.
Motowidlo, S. J., Dunnette, M. D., & Carter, G. W. (1990). An alternative selection procedure:
The low-fidelity simulation. Journal of Applied Psychology, 75, 640-647.

173

Mount, M. K., Sytsma, M. R., Hazucha, J. F., & Holt, K. E. (1997). Rater–ratee race effects in
developmental performance ratings of managers. Personnel Psychology, 50, 51-69.
Mumford, T. V., Campion, M. A., & Morgeson, F. P. (2006). Situational judgment in work
teams: A team role typology. In J. A. Weekley & R. E. Ployhart (Eds.), Situational
judgment tests: Theory, measurement, and application (pp. 319–343). Mahwah, NJ:
Erlbaum.
Mumford, T. V., Van Iddekinge, C. H., Morgeson, F. P., & Campion, M. A. (2008). The team
role test: Development and validation of a team role knowledge situational judgment test.
Journal of Applied Psychology, 93, 250-267.
Nelson, N. L., & Probst, T. M. (2004). Multiple minority individuals: Multiplying the risk of
workplace harassment and discrimination. In J. L. Chin (Ed.), The psychology of
prejudice and discrimination: Ethnicity and multiracial identity (Vol 2., pp. 193-217).
Westport, CT: Praeger Publishers/Greenwood Publishing Group, Inc.
Nguyen, N. T., & McDaniel, M. A. (2003). Response instructions and racial differences in a
situational judgment test. Applied H.R.M. Research, 8, 33-44.
Olson-Buchanan, J. B., Drasgow, F., Moberg, P. J., Mead, A. D., Keenan, P. A., & Donovan, M.
A. (1998). Interactive video assessment of conflict resolution skills. Personnel
Psychology, 51, 1-24.
Oppler, S. H., Campbell, J. P., Pulakos, E. D., & Borman, W. C. (1992). Three approaches to the
investigation of subgroup bias in performance measurement: Review, results, and
conclusions. Journal of Applied Psychology, 77, 201-217.
Organ, D. W. (1997). Organizational citizenships behavior: It’s construct cleanup time. Human
Performance, 10, 85-97.
Oyserman, D., Coon, H. M., & Kemmelmeier, M. (2002). Rethinking individualism and
collectivism: Evaluation of theoretical assumptions and meta-analyses. Psychological
Bulletin, 128, 3-72.
Peeters, H., & Lievens, F. (2005). Situational judgment tests and their predictiveness of college
students’ success: The influence of faking. Educational and Psychological Measurement,
65, 70-89.
Perkins, L. A., Thomas, K. M., & Taylor, G. A. (2000). Advertising and recruitment: Marketing
to minorities. Psychology & Marketing, 17, 235-255.
Ployhart, R. E., & Ehrhart, M. G. (2003). Be careful what you ask for: Effects of response
instructions on the construct validity and reliability of situational judgment tests.
International Journal of Selection and Assessment, 11, 1-16.

174

Podsakoff, P. M., MacKenzie, S. B., Paine, J. B., & Bachrach, D. G. (2000). Organizational
citizenship behaviors: A critical review of the theoretical and empirical literature and
suggestions for future research. Journal of Management, 26, 513-563.
Podsakoff, N. P., Whiting, S. W., Podsakoff, P. M., & Blume, B. D. (2009). Individual- and
organizational-level consequences of organizational citizenship behaviors: A metaanalysis. Journal of Applied Psychology, 94, 122-141.
Pulakos, E. D., White, L. A., Oppler, S. H., & Borman, W. C. (1989). Examination of race and
sex effects on performance ratings. Journal of Applied Psychology, 74, 770-780.
Reyna, C. (2008). Ian is intelligent but Leshaun is lazy: Antecedents and consequences of
attributional stereotypes in the classroom. European Journal of Psychology and
Education, 23, 439-458.
Richman-Hirsch, W. L., Olson-Buchanan, J. B., & Drasgow, F. (2000). Examining the impact of
administration medium on examinee perceptions and attitudes. Journal of Applied
Psychology, 85, 880-887.
Ridgeway, C. L. & Smith-Lovin, K. (1999). The gender system and interaction. Annual Review
of Sociology, 25, 191-216.
Riordan, C. M., & Shore, L. M. (1997). Demographic diversity and employee attitudes: An
empirical examination of relational demography within work units. Journal of Applied
Psychology, 82, 342-358.
Russell, A. & Owens, L. (1999). Peer estimates of school-aged boys' and girls' aggressionto
same- and cross-sex targets. Social Development, 8, 364-379.
Sackett, P. R., & DuBois, C. L. Z. (1991). Rater-ratee race effects on performance evaluation:
Challenging meta-analytic conclusions. Journal of Applied Psychology, 76, 873-877.
Saulny, S. (2011a, January 30). Black? White? Asian? More young Americans choose all of the
above. The New York Times, p. A1.
Saulny, S. (2011b, February 10). Counting by race can throw off some numbers. The New York
Times, p. A1.
Sagar, H. A., & Schofield, J. W. (1980). Racial and behavioral cues in Black and White
children’s perceptions of ambiguously aggressive acts. Journal of Personality and Social
Psychology, 39, 590-598.
Saucier, D. A., Miller, C. T., & Doucet, N. (2005). Differences in helping Whites and Blacks: A
meta-analysis. Personality and Social Psychology Review, 9, 2-16.
Schmitt, A.P. (1988). Language and cultural characteristics that explain differential item
functioning for Hispanic examinees on the Scholastic Aptitude Test. Journal of
Educational Measurement, 25, 1-13.
175

Shute, R., & Charlton, K. (2006). Anger or compromise? Adolescents’ conflict resolution
strategies in relation to gender and type of peer relationship. International Journal of
Adolescence and Youth, 13, 55-69.
Sidanius, J., & Pratto, F. (1999). Social dominance. Cambridge: Cambridge University Press.
Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self.
Advances in Experimental Social Psychology, 21, 261-302.
Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social categorization and
intergroup behaviour. European Journal of Social Psychology, 1, 149-178.
Thomas, D. A. (1990). The impact of race on managers’ experiences of developmental
relationships (mentoring and sponsorship): An intra-organizational study. Journal of
Organizational Behavior, 11, 479-492.
Thomas, K. W., Thomas, G. F., & Schaubhut, N. (2008). Conflict styles of men and women at
six organization levels. International Journal of Conflict Management, 19, 148-166.
Timberlake, J. M., & Estes, S. B. (2007). Do racial and ethnic stereotypes depend on the sex of
target group members? Evidence from a survey-based experiment. The Sociological
Quarterly, 48, 399-433.
Tjosvold, D. (1998). Cooperative and competitive goal approach to conflict: Accomplishments
and challenges. Applied Psychology: An International Review, 47, 285-313.
Truxillo, D. M., Bauer, T. N., & Sanchez, R. J. (2001). Multiple dimensions of procedural
justice: Longitudinal effects on selection system fairness and test-taking self-efficacy.
International Journal of Selection and Assessment,9, 336-349.
Tsui, A. S., Egan, T. D., & O’Reilly, C. A. III. (1992). Being different: Relational demography
and organizational attachment. Administrative Science Quarterly, 37, 549-579.
Tsui, A. S., & O’Reilly, C. A. III. (1989). Beyond simple demographic effects: Importance of
relational demography in superior-subordinate dyads. Academy of Management Journal,
32, 402-423.
Watson, W. E., Kumar, K., & Michaelsen, L. K. (1993). Cultural diversity’s impact on
interaction process and performance: Comparing homogenous and diverse task groups.
Academy of Management Journal, 36, 590-602.
Weekley, J. A., & Jones, C. (1997). Video-based situational testing. Personnel Psychology, 50,
25-49.
Wesolowski, M. A., & Mossholder, K. W. (1997). Relational demography in supervisorsubordinate dyads: Impact on subordinate job satisfaction, burnout, and perceived
procedural justice. Journal of Organizational Behavior, 18, 351-362.

176

Whetzel, D. L., & McDaniel, M. A. (2009). Situational judgment tests: An overview of current
research. Human Research Management Review, 19, 188-202.
Whetzel, D. L., McDaniel, M. A., & Nguyen, N. T. (2008). Subgroup differences in situational
judgment test performance: A meta-analysis. Human Performance, 21, 291-309.
Whittler, T. E. (1989). Viewers’ processing of actors’ race and message claims in advertising
stimuli. Psychology and Marketing, 6, 287-309.
Whittler, T. E. (1991). The effects of actors’ race in commercial advertising: Review and
extension. Journal of Advertising, 20, 54-60.
Whittler, T. E., & DiMeo, J. (1991). Viewers’ reactions to racial cues in advertising stimuli.
Journal of Advertising Research, 31, 37-46.
Williams, R. E., & Qualls, W. J. (1989). Middle-class Black consumers and intensity of ethnic
identification. Psychology and Marketing, 6, 263-286.
Wilson, L. K., & Gallois, C. (1985). Perceptions of assertive behavior: Sex combination, role
appropriateness, and message type. Sex Roles, 12, 125-141.
Wilson, K. L., Lizzio, A. J., Whicker, L., Gallois, C., & Price, J. (2003). Effective assertive
behavior in the workplace: Responding to unfair criticism. Journal of Applied Social
Psychology, 33, 362-395.
Zenger, T. R., & Lawrence, B. S. (1989). Organizational demography: The differential effects of
age and tenure distributions on technical communication. Academy of Management
Journal, 32, 353-376.

177